U.S. patent number 6,487,561 [Application Number 09/224,896] was granted by the patent office on 2002-11-26 for apparatus and methods for copying, backing up, and restoring data using a backup segment size larger than the storage block size.
This patent grant is currently assigned to EMC Corporation. Invention is credited to Dan Arnon, Zoran Cakeljic, Sharon Galtzur, Michael Hirsch, Peter Kamvysselis, Samuel Krikler, Yuval Ofek.
United States Patent |
6,487,561 |
Ofek , et al. |
November 26, 2002 |
Apparatus and methods for copying, backing up, and restoring data
using a backup segment size larger than the storage block size
Abstract
Method and apparatus for copying, transferring, backing up and
restoring data are disclosed. The data can be copied, backed up or
restored in segments sizes larger than the data blocks which
comprise a logical object. In some embodiments, the segment can
correspond to a track of a primary storage device and the data
blocks to a fixed size block. In some instances, copying, storage
and transfer of the segments which include multiple data blocks can
result in transfer of a data block not in a logical object.
Inventors: |
Ofek; Yuval (Framingham,
MA), Cakeljic; Zoran (Newton, MA), Krikler; Samuel
(Ramat HaSharon, IL), Galtzur; Sharon (Holon,
IL), Hirsch; Michael (Mazkeret Batya, IL),
Arnon; Dan (Boston, MA), Kamvysselis; Peter (Boston,
MA) |
Assignee: |
EMC Corporation (Hopkinton,
MA)
|
Family
ID: |
22842673 |
Appl.
No.: |
09/224,896 |
Filed: |
December 31, 1998 |
Current U.S.
Class: |
1/1; 707/999.204;
707/999.202; 707/E17.01; 714/E11.125; 714/E11.123; 707/999.2;
707/999.201; 711/162; 714/20; 714/17; 714/15; 712/228 |
Current CPC
Class: |
G06F
11/1451 (20130101); G06F 11/1464 (20130101); G06F
16/10 (20190101); Y10S 707/99953 (20130101); Y10S
707/99955 (20130101); Y10S 707/99952 (20130101) |
Current International
Class: |
G06F
17/30 (20060101); G06F 017/30 (); G06F 012/00 ();
G06F 015/00 (); H02H 003/05 () |
Field of
Search: |
;707/204,202
;714/15,17,20 ;711/162 ;712/228 |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
Other References
US. patent application Ser. No. 09/224,637, Ofek et al., filed Dec.
31, 1998. .
U.S. patent application Ser. No. 09/224,638, Ofek et al., filed
Dec. 31, 1998. .
U.S. patent application Ser. No. 09/224,897, Ofek et al., filed
Dec. 31, 1998. .
U.S. patent application Ser. No. 09/223,896, Ofek et al., filed
Dec. 31, 1998. .
U.S. patent application Ser. No. 09/223,897, Ofek et al., filed
Dec. 31, 1998..
|
Primary Examiner: Coby; Frantz
Attorney, Agent or Firm: Wolf, Greenfield & Sacks,
P.C.
Claims
What is claimed is:
1. A method of copying a logical object, the logical object being
stored in a plurality of physical blocks of a computer storage
device, the physical blocks being stored in storage segments of the
computer storage device, the method comprising steps of:
identifying a set of the storage segments, each storage segment of
the set including at least one of the physical blocks of the
logical object; and copying the identified storage segments;
wherein at least one of the copied storage segments includes a data
block that is not a part of the logical object; and wherein at
least one of the identified storage segments includes data from a
physical block that does not include data for the logical
object.
2. The method of claim 1, wherein each storage segment corresponds
to a track of a disk in the computer storage device.
3. The method of claim 2, wherein the physical blocks are fixed
size data blocks.
4. The method of claim 1, wherein the step of copying comprises a
step of writing the identified storage segments to one or more
storage tapes.
5. The method of claim 1, wherein the step of copying comprises a
step of writing the identified storage segments to one or more
storage tapes.
6. The method of claim 1, wherein the step of copying comprises a
step of writing the identified storage segments to a secondary
storage device.
7. A method of copying a logical object, the logical object being
stored in a plurality of physical blocks of a computer storage
device, the physical blocks being stored in storage segments of the
computer storage device, the method comprising steps of: receiving
a set of the storage segments, at least one of the stored storage
segments including a plurality of physical data blocks, and each
storage segment of the set including at least one of the physical
blocks of the logical object; and storing the received storage
segments without separately addressing the physical data blocks of
the storage segments.
8. The method of claim 7, wherein each storage segment corresponds
to a track of a disk in the computer storage device.
9. The method of claim 8, wherein the physical blocks are fixed
size data blocks.
10. The method of claim 7, wherein the step of storing comprises a
step of writing the received storage segments to one or more
storage tapes.
11. The method of claim 7, wherein at least one of the received
storage segments includes a data block that does not include data
for the logical object.
12. The method of claim 11, wherein the step of storing comprises a
step of writing the received storage segments to one or more
storage tapes.
13. The method of claim 11, wherein the step of storing comprises a
step of writing the received storage segments to a secondary
storage device.
14. A computer readable media storing a back up copy of a logical
object, the logical object including a plurality of data blocks,
comprising: a plurality of data segments stored on the readable
media, each data segment including at least one of the logical
object data blocks, at least one of the data segments including a
plurality of the logical object data blocks, and at least one of
the data segments including a data block that does not include any
data for the logical object.
15. The media of claim 14, wherein each data segment corresponds to
a track of a disk in a primary storage element storing the logical
object.
16. The media of claim 14, further comprising: a metadata segment,
stored on the readable media, to identify the data blocks of the
data segments which are in the logical object.
17. The media of claim 16, wherein the data segments and metadata
form an abstract block set.
18. The media of claim 17, wherein the metadata comprises a
plurality of labels, each label associated with one or more of the
data segments and a table associating the labels with a relative
position in the logical object.
19. The method of claim 17, wherein the metadata comprises physical
memory addresses corresponding to the location of logical data
blocks of the logical object stored in a primary storage
device.
20. The method of claim 19, wherein the metadata comprises a
physical address associated with an extent of storage segments, and
a table specifying the relative position of the extents in the
logical object.
21. A computer storage system, comprising: a computer storage
device including a plurality of physical storage segments, each to
store at least one data block; means for identifying a set of the
storage segments, wherein at least one of the identified storage
segments includes a plurality of physical data blocks, each storage
segment of the set including at least one physical block of a
logical object; and means for transmitting the logical object by
transmitting the identified storage segments without separately
addressing each of the physical data blocks.
Description
FIELD OF THE INVENTION
This invention relates to data storage for computers, and more
particularly to an apparatus and methods for copying, backing up,
and restoring data using a backup segment size larger than the
storage block size.
DISCUSSION OF THE RELATED ART
Virtually all computer applications (or programs) rely on storage.
This storage can be used for both storing the computer code and for
storing data manipulated by the code. (The term "data" refers to
any information, including formatting information, executable code
and data for manipulation by an application program.)
Storage technology has developed in a variety of different
directions. Accordingly, a wide variety of storage systems are
available. It has become impractical, therefore, for the person
writing the computer application to also be responsible for
detailed control over how data is stored on the storage system.
For this (and other) reasons, application programs typically run on
an operating system (e.g., Unix, Windows, MS DOS, Linux, and the
many variations of each). Once again, however, the operating system
may be used with a variety of storage systems.
It would be highly inefficient to have to change the operating
system, or the application programs, every time a change is made to
physical storage. As a result, various layers of abstraction have
evolved for viewing how data is actually stored in the storage
system.
FIG. 1 illustrates one way of viewing the layers of abstraction. At
the top level 10, the application program may assume that data is
stored in a manner that has very little to do with how the data is
placed onto the physical device. For example, the application may
view the storage system as containing a number of directories and
data files within the directories. Thus, in an application written
for use in the Unix operating system, the application will assume
that files are stored according to the Unix directory structure
(including hierarchical directories and files located within the
directories). This assumed organization of physical storage may
have very little to do with how that data is actually stored onto
the actual storage devices. This view may be referred to as the
"logical view" because of the separation between the logical view
of data from the application level is divorced from any view of how
the data is physically stored. A logical entity, such as a file,
database or other construct, may be referred to at the logical
level as a "logical object."
The application level 10 interfaces with the file system level 12.
The file system level is concerned with how files are stored on
disks and how to make everything work efficiently and reliably.
Thus, the file system level may be responsible for storing
directory structure, and for breaking up files into constituent
data blocks for storage onto a physical storage system. For
example, in most implementations of Unix, each file has an
associated I-node. This node may contain accounting and protection
information and, additionally, a set of pointers to data
blocks.
Relatively early in the development of computer systems, disk
drives became a fundamental device for storage. Accordingly,
computer operating systems have been developed assuming that memory
will rely on input/output ("I/O") to a disk drive. The file system
12, therefore, may assume one or more "volumes" which correspond to
a physical storage unit such as a disk drive (or any other unit of
storage), with data stored in blocks on the disk drive.
The demand for storage to be available for use by applications has
sky rocketed. As a result, a number of separate physical devices
may be required to accommodate the total amount of storage required
for a system. In addition, storage systems are often changed or
reconfigured.
To insulate the operating system from any changes within the
physical device storage system, some mechanism is often employed to
flexibly map a standard (volume) view of physical storage onto an
actual physical storage system. The logical volume manager ("LVM")
14 of FIG. 1 can help achieve this function by mapping the file
system view of data storage into an intermediate layer.
Finally, the actual storage reading and writing (and, potentially,
additional mapping onto physical storage devices) occurs within the
physical storage system level 16, as illustrated in FIG. 1. Thus,
for example, the logical volume manager may map the file system
level view of data into volume sizes corresponding to fixed
physical storage segment sizes for storage on a physical device
(e.g, block sizes). The physical storage system level may then map
the logical volume manager level volumes onto physical storage
segments (e.g., hyper-volumes discussed below).
Logical volume managers have been implemented for use with the
HP-UX by HP and by VERITAS operating systems, as examples. The
Symmetrix line of storage systems, available from EMC Corporation,
of Hopkinton, Mass., is one system capable of mapping hyper-volumes
onto physical devices. (The Symmetrix product line of integrated
cached disk arrays is described in numerous publications form EMC
Corporation, including the Symmetrix model 55xx product manual,
p-n200-810-550, rev.f, February, 1996.)
In the above examples, the mapping of application level data into
actual physical storage occurs across four levels: application
level to file system level; file system level to LVM level; LVM
level to physical storage system level; and physical storage system
level to the actual physical storage devices. More or fewer levels
of mapping can be done. In some systems, for example, only one
level of mapping is performed, e.g., mapping from the application
level directly onto actual physical storage devices. In many
systems, the mapping stage at the LVM level is omitted. Similarly,
in many systems, no mapping is done at the physical storage level
(e.g., data is stored directly onto actual devices corresponding to
the format of the preceding level and without any further mapping
onto physical storage components.)
FIG. 2A illustrates an example of the mapping that may be performed
by the logical volume manager 14 and the physical storage system
16, to store data onto actual physical devices. The
application/file system's view of the storage system contemplates
three separate storage devices--volume A 20, volume B 21, and
volume C 22. Thus, as far as the file system level 12 can discern,
the system consists of three separate storage devices 20-22. Each
separate storage device may be referred to as a "virtual volume,"
or "virtual disk." This reflects that the operating system's view
of the storage device structure may not correspond to the actual
physical storage system implementing the structure (hence,
"virtual"). Unlike the application level 10, however, the file
system 12 perspective is as if the file system 12 were dealing with
raw physical devices or volumes.
As far as the file system level is concerned, the virtual volumes
may be divided up into "partitions," which are continuous segments
of storage. These partitions are, in fact, "virtual" partitions,
because the partition may actually be stored across a variety of
physical storage segments (e.g., hyper-volumes).
In FIG. 2A, the data is physically stored on the physical storage
devices 24-26. In this particular example, although there are three
physical devices 24-26 and three volumes 20-22, there is not a one
to one mapping of the virtual volumes to physical devices. In this
particular example, the data in volume A 20 is actually stored on
physical devices 24-26, as indicated at 20a, 20b and 20c. In this
example, volume B is stored entirely on physical device 24, as
indicated at 22a, 22b. Finally, volume C is stored on physical
device 24 and physical device 26 as indicated at 21a, 21b.
In this particular example, the boxes 20a-20c, 21a-21b and 22a-22b
represent contiguous segments of storage within the respective
physical devices 24-26. These contiguous segments of storage may,
but need not, be of the same size. The segments of storage may be
referred to as "hyper-volumes," and correspond to segments of
physical storage that can be used as components when constructing a
virtual volume for use by the file system. A hypervolume may be
comprised of a number of "data blocks." A data block is a unit of
storage (e.g., a 512 byte block) that is written or read at one
time from the physical storage device.
Array management software running on a general purpose processor
(or some other mechanism such as a custom hardware circuit) 23
translates requests from a host computer (not shown) (made assuming
the logical volume structure 20-22) into requests that correspond
to the way in which the data is actually stored on the physical
devices 24-26. In practice, the array management software 23 may be
implemented as a part of a unitary storage system that includes the
physical devices 24-26, may be implemented on a host computer, or
may be done in some other manner.
In FIG. 2A the array management software 23 performs the functions
of both the logical volume manager 14 (if present) and the physical
storage level 16, by mapping the file system's virtual volumes
20-22 into segments that can be stored onto physical devices 24-26.
The array management software 23 also performs the functions of the
physical storage system level 16, by determining where to store the
hyper-volumes 20A-20C, 21A-21B and 22A-22B.
The physical storage devices shown in the example of FIG. 2A are
disk drives. A disk drive may include one or more disks of a
recording media (such as a magnetic recording medium or an optical
recording medium). Information can be written and read from this
storage medium for storage purposes. The recording medium is
typically in the form of a disk that rotates. The disk generally
includes a number of tracks on which the information is recorded
and from which the information is read. Each track may include more
than one "data block." A data block is a unit of data that can be
read as a single unit. A data block may be a 512 by the block of
data, an 8 k segment on a 32 k track, or some other structure. In
these examples, the size of the block is fixed. In other cases, the
block may be of variable size, such as a CKD record. In a disk
drive that includes multiple disks, the disks are conventionally
stacked so that corresponding tracks of each disk overlie each
other. In this case, specification of a single track on which
information is stored within the disk drive includes not only
specification of an individual track on a disk, but also which of
the multiple disks the information is stored on.
To identify an individual data block, an address may include a
specification of the disk, (which may consist of several
"platters"), a specification of the track within the disk (or
"cylinder"), a specification of the head (or which of the platters
comprising the "disk") and a specification of the particular data
block within the track. The specification of the position of the
data block within the track may, for example, be addressed as an
offset, e.g., this is the third data block appearing on the track.
Thus, an address of ddcccch:offset may specify a block--disk dd,
cylinder cccc, head h and the specified offset. The physical
storage devices for use with the present invention may, however, be
formed in any other geometry, addressed in any other manner or even
constitute a different type of storage mechanism.
FIG. 2B illustrates one example of mapping between the top level of
abstraction--the application level--to the actual physical storage
level. An application level file 200 includes visual information.
This information is in the form of a conventional file and includes
a series of bits.
When the application level file is mapped onto physical storage,
the application level file may be converted into segments of the
individual bits, e.g., segment 203. Thus, a segment of the
application level file 203 is mapped (for example according to the
general mapping structure described above with reference to FIG. 1)
onto actual physical storage devices 204-206. In this example, the
first segment of bits in 203 in the application level file 200 is
mapped onto physical storage device 204, at a portion 208 of the
physical storage device 204. As shown in FIG. 2B, the individual
segments of bits in the application level file 200 may be mapped
anywhere among a plurality of actual physical storage devices. The
granularity of the segments of bits (e.g., segment 203) may
correspond to one of a variety of different levels. For example,
the granularity of the segments may be a 512 byte data block. In
another embodiment, the granularity may correspond to the amount of
data stored in a track of the physical storage device 204-206 (when
the physical storage devices are disk drives).
FIG. 2C illustrates an example of a logical object 27 that includes
six data blocks or logical block elements 27a-27f. The logical
object itself may be any data structure or collection of data. For
example, the logical object could be a database table, a portion of
a file system file, or a complete file system file, or any other
identifiable logical object. Each of the data blocks 27a-27f may be
a fixed size data block, or a varying size data block such as a CKD
record.
In the example of FIG. 2C, the logical object is stored on a
physical storage device 28. In this example, the storage device
includes a number of columns, each representing a track of a
disk.
Each row of the physical storage device represents a physical data
or block element 15 within the applicable column/track. For
example, row 28a, column 28b, stores a data block corresponding to
the logical block element 27b. Track 28b would store physical data
blocks that have the contents of logical block elements 27a and
27b. As can be seen from FIG. 2C, the logical block elements can be
stored in any order on the physical devices.
While the physical storage device 28 is illustrated as a contiguous
array, this need not be the case. For example, each of the tracks,
such as column 28b, may be stored on a different disk drive or be
part of a different hypervolume.
In a system including an array of physical disk devices, such as
disk devices 24-26 of FIG. 2A, each device typically performs error
detection and/or correction for the data stored on the particular
physical device. Accordingly, each individual physical disk device
detects when it does not have valid data to provide and, where
possible, corrects the errors. Even where error correction is
permitted for data stored on the physical device, however, a
catastrophic failure of the device would result in the
irrecoverable loss of data.
Accordingly, storage systems have been designed which include
redundant storage capacity. A variety of ways of storing data onto
the disks in a manner that would permit recovery have developed. A
number of such methods are generally described in the RAIDbook, A
Source Book For Disk Array Technology, published by the RAID
Advisory Board, St. Peter, Minn. (5th Ed., February, 1996). These
systems include "RAID" storage systems. RAID stands for Redundant
Array of Independent Disks.
FIG. 3A illustrates one technique for storing redundant information
in a RAID system. Under this technique, a plurality of physical
devices 31-33 include identical copies of the data. Thus, the data
M1 can be "mirrored" onto a portion 31a of physical device 31, a
portion 32a of physical device 32 and a portion 33a of physical
device 33. In this case, the aggregate portions of the physical
disks that store the duplicated data 31a, 32a and 33a may be
referred to as a "mirror group." The number of places in which the
data M1 is mirrored is generally selected depending on the desired
level of security against irrecoverable loss of data.
In a mirror group, the copies are "linked." That is, any update to
one mirror causes an update to each other mirror in the group.
FIG. 3A shows three physical devices 31-33 which appear to be
located in close proximity, for example within a single storage
system unit. For very sensitive data, however, one or more of the
physical devices that hold the mirrored data may be located at a
remote facility.
"RAID 1" is an example of data redundancy through mirroring of
data. In a RAID 1 architecture, a number of different mechanisms
may be used for determining how to access and update data to
improve, for example, performance of the storage system. In any
event, a RAID 1 architecture certainly has the ability to recover
lost data. Unfortunately, the RAID 1 architecture multiplies the
cost of physical storage by the number of "mirrors" included in the
mirror group.
FIG. 3B illustrates a solution that requires less added storage. In
FIG. 3B, data is stored at locations 34a-34d. In this particular
example, the physical device 33 includes parity information P1 at
35a, 35b. The parity information is generated by a simple
exclusive-OR ("XOR") of the corresponding bits of data. Thus, the
parity information P1 would be generated by XORing the
corresponding bits of the data D1 and data D2.
A variety of mechanisms are known for distributing the parity
information on the physical devices. In the example shown in FIG.
3B, all of the parity information is stored on a single physical
device 33. In other cases, the parity information may be
distributed across the physical devices.
FIG. 4 illustrates the concept that, within a given disk array,
there is no need for all of the data to follow the same redundancy
rule. In FIG. 4, a first group of storage segments on physical
devices 40-42 form a mirror group 44. In the mirror group 44, the
entire contents of a single logical volume (HV-A) are mirrored on
three different physical devices 40-42.
In FIG. 4, a single virtual volume is stored on the fourth physical
device 43, without any redundancy information, as indicated at
46.
Finally, a last group of data segments 45, on all four physical
devices 40-43, implement a parity redundancy scheme. In this
particular example, the parity information is stored in segments of
memory on two different physical devices 42-43, as indicated at 47a
and 47b.
The storage system of FIG. 4 contains redundant information that
permits recovery from errors, including use of a mirror for data
located at a remote facility, that also permits recoveries from
catastrophic failure.
FIG. 5 illustrates one system for additional backup, which may be
used or adapted in accordance with certain aspects of the present
invention. In FIG. 5, a computer or client 50 performs its
operations using storage system 52. The client 50 may be any
conventional computing system, such as a network client available
from Sun Microsystems, and running the Solaris operating system (a
version of Unix), an HP client running HP-UX (a Hewlett-Packard
client, running a Hewlett-Packard version of the Unix operating
system) or an IBM client running the AIX operating system (an IBM
version of Unix) or any other system with an associated operating
system. The storage system 52 may be any conventional storage
system, including a Symmetrix storage system, described above. The
client 50 may be connected to many other devices over a network
56.
A backup storage system 54 is also attached to the network 56. The
backup storage system 54 includes a backup storage device (which
may be disk drives, tape storage or any other storage mechanism),
together with a system for placing data into the storage and
recovering the data from that storage.
To perform a backup, the client 50 copies data from the storage
system 52 across the network 56 to the backup storage system 54.
This process can be explained in greater detail with reference to
FIG. 1. The storage system 52 may correspond to the actual physical
storage 16 of FIG. 1. For the client 50 to write the backup data
over the network 56 to the backup storage system 54, the client 50
first converts the backup data into file data--i.e. gets the data
from the physical storage system level 16, and converts the data
into application level format (e.g. a file) through the logical
volume manager level 14, the file system level 12 and the
application level 10. Thus, an actual data file may be communicated
over the network 56 to the backup storage device 54. When the
backup storage device 54 receives the data file, the backup storage
system 54 can take the application level 10 data file, convert it
to its appropriate file system level 12 format for the backup
storage system, which can then be converted through a logical
volume manager 14 level and into physical storage 16.
This form of backing up data may be referred to as
"logical--logical" backup. That is, the logical data is backed up
on the backup storage device 54. The data to be backed up is
presented independent of the manner in which it is physically
stored on storage system 52 at the physical storage system level
16, independent of the file system level mechanisms on the client
50, and independent of how data is stored on the backup storage
device 54.
The EDM (EMC Data Manager) line of products is capable of
logical-logical backup over a network, as described in numerous
publications available from EMC, including the EDM User Guide
(Network) "Basic EDM Manual".
FIG. 6 illustrates one embodiment of an alternative structure for
backup of data which may also be used in accordance with the
present invention. In the embodiment of FIG. 6, a direct connection
60 is established between the storage system 52 and the backup
storage system 54. In this embodiment, the backup storage system
may be a system as generally described in EMC Data Manager:
Symmetrix Connect User Guide, P/N 200-113-591, Rev. C, December
1997, available from EMC Corporation of Hopkinton, Mass. The direct
connection 60 may be a high speed data channel, such as a SCSI
cable or one or more fiber-channel cables. In this system, a user
may be permitted to backup data over the network 56, or the direct
connection 60.
While the method and apparatus of the present invention may be
described with reference to the systems and concepts described
above and in the discussion of the related art, this is not
intended to be limiting. The present invention has broader
application. Certain aspects of the invention may be applied to any
storage system. Accordingly, the invention is only limited by the
claims set forth below.
Whether the restore and backup process is done at a logical level
or at a physical level, backups in the prior art require copying a
complete file (or in some instances even more, such as an entire
partition) for the backup. Methods of backing up and restoring data
on the system of FIG. 6 are described in co-pending and commonly
owned U.S. patent application Ser. No. 09/052,579, entitled
"Logical Restore From A Physical Backup In A Computer Storage
System," filed Mar. 31, 1998, and naming John Deshayes and Madhav
Mutalik as inventors, and which is hereby incorporated herein by
reference in its entirety.
FIG. 7 shows a storage system 70 that may be used as the storage
system 52 of FIG. 6. The client 50 may be connected to the storage
device using a channel or bus 71. The channel for communication
with the client 50 can be any suitable connection such as a Small
Computer System Interface ("SCSI") or Enterprise Systems Connection
Architecture ("ESCON"). While only one communication channel 71
into the storage system 70 is shown in FIG. 7, other channels may
be included. (While the method and apparatus of the present
invention may be described with reference to the storage system of
FIG. 6 and the physical storage system (and associated features and
methods) of FIG. 7, this is not intended to be limiting. The
present invention has broader application. Certain aspects of the
invention may be applied to any storage system.)
Within the storage system 70 is a host adapter 72. In this
particular embodiment, the host adapter 72 is responsible for
managing and translating read and write requests from the host
computer (e.g., client 52 or backup storage system 54), which are
based on the virtual disk structure (e.g., from the file system or
logical volume manager level), into one or more requests
corresponding to how data is stored on the actual physical storage
devices 76a-76d of the storage system 70. Thus, in this embodiment,
the host adapter 72 implements at least some of the array
management software 23 functions of FIG. 2. The host adapter 72 can
be implemented in any of a number of ways, including using a
general purpose processor or a custom hardware implementation. In
addition, multiple host adapters may be included to facilitate
having additional I/O channels for the storage system 70.
The host adapter 72 communicates with the other components of the
storage system 70 using bus 73. The bus 73 may be any suitable
communication element, including use of SCSI, ESCON, and other bus
protocols.
Access to the physical storage devices 76a-76d is controlled
through the use of disk adapters 75a-75d. The disk adapter 75a-75d
can also be implemented using a general purpose processor or custom
hardware design. In the embodiment illustrated in FIG. 7, a disk
adapter is provided for each physical storage device. A disk
adapter can, of course, have more than one storage device attached
to it. In addition, disk adapters may include secondary connections
to the physical storage devices of another disk adapter. This
permits recovery from failure of one disk adapter by shifting its
functions to the second disk adapter.
In the embodiment of FIG. 7, reading and writing to the physical
storage device 76a-76d through the disk adapters 75a-75d is
facilitated through use of a cache 74. The cache 74 may be a random
access memory having greater speed than the disk drives. When
reading data, if the data is being temporarily stored in the cache,
the read request can be fulfilled more quickly by taking the data
from the cache 74. Similarly, when writing data, the data to be
written can be stored in the cache. The other components of the
system can proceed, while the data is written from the cache to the
applicable physical storage device.
Any of a variety of mechanisms can be used to implement and manage
the cache. An example of such a mechanism is included in U.S. Pat.
No, 5,537,568, entitled "System for dynamically controlling cache
manager maintaining cache index and controlling sequential data
access," issued on Jul. 16, 1996. Similarly, writes may be
accomplished through the cache using any of a variety of mechanisms
and strategies. One mechanism for writing from the cache is to
store the data to be written in the cache, and mark a "write
pending" bit. When the write pending bit is encountered, the
applicable data can be written to the disk. This technique is
described generally in U.S. Pat. No. 5,341,493, entitled "Disk
storage system with write preservation during power failure,"
issued on Aug. 23, 1994.
The cache may be divided into more than one area. For example, the
cache may include an area 74a for storing data being read or
written from physical storage devices 76a-76d. The cache may
further include a "mailbox" area 74b. The mailbox area 74b may be
used to facilitate communications among the disk adapters 75a-75d
and with the host adapter 72. For example, each disk adapter may
have its own area within the mailbox 74b. Each of the disk adapters
75a-75d can post or read information from the applicable mailbox
area 74b, to communicate status and other information.
A remote adapter 78 may also be attached to the bus 73 of the
storage system 70. The remote adapter may be employed for
communication with remote data facilities ("RDF"), for example,
connection to another storage device to maintain a mirror
redundancy group. One form of RDF link and method of implementation
is described in various publications available from EMC
Corporation, including SYMMETRIX Remote Data Facility Product
Manual, P/N 200-999-554, rev. B, June 1995. RDF embodiments are
also described in U.S. Pat. No. 5,544,347 (Yanai) which is hereby
incorporated herein by reference in its entirety. It should be
appreciated, however, that the present invention is not limited to
the use of RDF or to a system that employs SYMMETRIX disk arrays,
and can be employed with any of numerous other types of storage
systems.
A service processor 77 may be coupled to the bus 73 of the storage
system 70. The service processor 77 may include a display, keyboard
and other I/O devices to permit an operator to use the service
processor 77 for configuring the components of the storage system
70 and for running or initiating diagnosis and maintenance
facilities.
SUMMARY OF THE INVENTION
According to one embodiment of the present invention, a computer
system is disclosed. According to this embodiment, the computer
system includes a host domain that has at least one host computer.
The computer system also includes a storage domain, coupled to the
host domain, that comprises a plurality of primary storage devices,
a secondary storage device and a switched network coupled to the
primary storage nodes and to the secondary storage node.
According to another embodiment of the present invention, a
computer system is disclosed that includes a plurality of host
computers, each of the host computers constituting a different
platform. The computer system further includes a plurality of
primary storage devices, each being associated with at least one of
the host computers. The system also includes a secondary storage
device, coupled to a plurality of the primary storage devices, the
secondary storage device being configured to receive backup data
from each of the host computers.
According to another embodiment of the present invention, a method
of transferring data from a primary storage node to a secondary
storage node is disclosed. According to this embodiment, a
connection is automatically established from one of the primary
storage elements to a secondary storage element, for transferring
data to the secondary storage element. Data is transferred from the
primary storage element directly to the secondary storage element
over the first connection.
According to another embodiment of the present invention, a method
of sending a copy of data from a storage element of a computer
system is disclosed. According to this embodiment, the data is
first formulated into an abstract block set. The abstract block set
is transmitted. In this and other embodiments, the steps of
formulating and transmitting may be performed sequentially or
concurrently.
According to another embodiment of the present invention, a method
of storing a logical object is disclosed. According to this
embodiment, the logical object is formulated into an abstract block
set and stored.
According to another embodiment of the present invention, a storage
device is disclosed. According to this embodiment, the storage
device includes a memory and means for transmitting an abstract
block set from the memory.
According to another embodiment of the present invention, a
secondary storage system is disclosed. According to this
embodiment, the secondary storage system includes a secondary
storage media and means for storing an abstract block set on the
secondary storage media.
According to another embodiment of the present invention, a
computer readable media storing a logical object is disclosed.
According to this embodiment, the media includes a plurality of
data blocks, each storing on the readable media a portion of data
from the logical object, and a metadata segment, stored on the
readable media, to identify the order of data blocks in the logical
object.
According to another embodiment of the present invention, a method
of generating a backup for a logical object is disclosed. According
to this embodiment, data blocks of the logical object that have
changed since an earlier point in time are identified. The
identified data blocks are stored as a differential abstract block
set.
According to another embodiment of the present invention, a storage
device is disclosed. According to this embodiment, the storage
device includes a memory, means for identifying data blocks that
have changed since an earlier point in time and means for
transmitting a differential abstract block set from the memory.
According to another embodiment of the present invention, a method
of forming an updated abstract block set is disclosed. According to
this embodiment, a full abstract block set is provided. A
differential abstract block set is also provided. The full abstract
block set and the differential abstract block set are combined to
form the updated abstract block set.
According to another embodiment of the present invention, a method
of forming an updated backup of a logical object is disclosed.
According to this embodiment, a first backup of the logical object
is provided. A differential backup of the logical object is also
provided, the differential backup including a plurality of backup
data blocks that have changed since the first backup was formed.
The backup data blocks are added to the first backup and metadata
identifying an order of data blocks in the updated backup is
added.
According to another embodiment of the present invention, a
secondary storage device is disclosed. According to this
embodiment, the storage device includes a secondary storage media
and a controller programmed to combine a first backup and a
differential abstract block set to form a full abstract block
set.
According to another embodiment of the present invention, a method
of copying a logical object is disclosed. According to this
embodiment, a set of storage segments of a computer storage device
are identified, each of the identified segments including data from
at least one physical block of a logical object. The identified
storage segments are copied. According to this embodiment, at least
one of the copied storage segments includes a plurality of the
physical data blocks. Thus, the size of the storage segment is not
necessarily the same as the size of individual physical data
blocks.
According to another embodiment of the present invention, a method
of creating a backup of a logical object is disclosed. According to
this embodiment, a set of backup segments is received, each backup
segment including at least one physical block of a logical object.
The received storage elements are stored, at least one of the
storage segments including a plurality of the physical data
blocks.
According to another embodiment of the present invention, a
computer readable media storing a backup copy of a logical object
is disclosed. According to this embodiment, a plurality of data
segments are stored on the readable media, each data segment
including at least one datablock of the logical object, and at
least one of the data segments including a plurality of the logical
data blocks. This embodiment further includes a metadata segment,
stored on the readable media, to identify data blocks of the
logical object in the data segments. In this embodiment, the data
segment may, for example, be a track including a plurality of fixed
size blocks.
According to another embodiment of the present invention, a
computer storage system is disclosed. According to this embodiment,
the system includes a computer storage device that includes a
plurality of physical storage segments (which, in one embodiment,
is a track) each storing at least one datablock. The system further
includes means for identifying a set of storage elements, each
storage segment of the set including at least one physical block of
a logical object and means for transmitting the identified storage
segments.
According to another embodiment of the present invention, a method
of backing up a logical object at a fixed point in time is
disclosed. According to this embodiment, a set of storage segments
that include logical data blocks of the logical object are
identified. These storage segments are copied to a backup storage
device, out of order from the order of storage segments or logical
data blocks appearing in the logical object. During the copying
step, if a storage segment that includes a physical block of the
logical object is to be modified, that storage segment is
immediately backed up. In this and other embodiments, the storage
segments may (but need not) correspond in size to the size of data
blocks.
According to another embodiment of the present invention, a
computer storage system is disclosed. According to this embodiment,
the system includes a computer storage device that has a plurality
of storage segments. The system further includes means for
identifying a set of the storage segments that includes logical
objects, logical data blocks; means for copying the identified
storage segments, out of order from the order of logical data
blocks and the logical object; and means for immediately copying
storage segments to the backup storage device if an attempt is made
to modify a physical block of the storage segment.
According to another embodiment of the present invention, a method
of copying a logical object to a primary storage device is
disclosed. According to this embodiment, a copy of the logical
object is provided. Physical blocks of memory in the primary
storage device are allocated for storing the logical object. A map
of the data blocks of the copy of the logical object to the
physical blocks of the primary storage device is created. The data
blocks are copied to the physical blocks, based on the map.
According to another embodiment of the present invention, a method
of copying a logical object to a primary storage device is
disclosed. According to this embodiment, an abstract block set copy
of the logical object is provided. Physical blocks of memory are
allocated in the primary storage device to store the logical
object. The data blocks of the copy of the logical object are
mapped to the physical blocks of the primary storage device and the
data blocks are copied to the physical blocks based on the
mapping.
According to another embodiment of the present invention, a
computer storage device is disclosed. According to this embodiment,
the device includes a memory including a plurality of physical data
blocks. The device further includes means for storing the data
blocks of an abstract block set to the physical data blocks, based
on a mapping of the data blocks to a set of the physical data
blocks.
According to another embodiment of the present invention, a method
of copying a logical object is disclosed. According to this
embodiment, a set of storage segments that includes the logical
data blocks are identified. The storage segments may correspond to
the logical data blocks, or may be of a different size. The
identified storage segments are copied to a second storage device,
out of order from the order of logical data blocks in the logical
object.
According to another embodiment of the present invention, a method
of copying a logical object is disclosed. According to this
embodiment, a set of storage segments that includes the logical
data blocks of the logical object are identified. The identified
storage segments are copied to a second computer storage device in
parallel. Metadata is provided to identify the order of data stored
in the identified storage segments in the logical object.
According to another embodiment of the present invention, a method
of backing up a logical object that includes a plurality of logical
blocks is disclosed. According to this embodiment, a first and a
second backup media are provided. In one embodiment, each backup
media is a digital storage tape. Logical blocks are written to the
first and the second backup media in parallel.
According to another embodiment of the present invention, a
secondary storage device is disclosed. According to this
embodiment, the secondary storage device includes a plurality of
storage components and means for writing portions of an abstract
block set to the storage components, in parallel.
According to another embodiment of the present invention, a method
of restoring a logical object is disclosed. According to this
embodiment, a first and a second portion of a copy of the logical
object are provided. Data blocks stored in the first portion and
data blocks stored in the second portion are read in parallel. The
logical object is restored from the read data blocks.
According to another embodiment of the present invention, a
secondary storage device is disclosed. According to this
embodiment, the secondary storage device includes means for reading
data from a plurality of storage components, in parallel, and means
for providing the read data to another device as an abstract block
set.
Each of the above disclosed inventions and embodiments may be
useful and applied separately and independently, or may be applied
in combination. Description of one aspect of the inventions are not
intended to be limiting with respect to other aspects of the
inventions.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 illustrates an example of conversion of application level
data to storage in a physical system and vice versa.
FIG. 2A illustrates an example of the relationship between logical
volumes and physical storage devices.
FIG. 2B illustrates an example of mapping a logical file onto a
physical storage system.
FIG. 2C illustrates another example of mapping a logical object
onto a physical storage device.
FIG. 3A illustrates an example of mirroring on different physical
storage devices.
FIG. 3B illustrates an example of redundant parity information on
physical storage devices.
FIG. 4 illustrates an example of multiple redundancy groups within
a single array of storage devices.
FIG. 5 illustrates an example of a backup storage system.
FIG. 6 illustrates one embodiment of a backup storage system that
includes a mechanism for direct backup of data on the primary
storage system.
FIG. 7 illustrates an example of a storage system.
FIG. 8 illustrates one example of a computer storage system
structured to have an enterprise host domain and an enterprise
storage domain or network, according to one embodiment of the
present invention.
FIG. 9 illustrates another example of a computer system including
an enterprise host domain and an enterprise storage domain,
according to one embodiment of the present invention.
FIG. 10 illustrates one embodiment of a method of copying or
backing up a logical object.
FIG. 11A illustrates one example of some of the components of a
computer system that includes a host domain and a storage domain,
according to one embodiment of the present invention.
FIG. 11B illustrates another embodiment of components of a computer
system that is divided into a host domain and a storage domain,
according to one embodiment of the present invention.
FIG. 12 illustrates one embodiment of a method for copying or
backing up data in a computer storage system.
FIG. 13 illustrates one example of mapping a logical object onto a
physical storage device and formation of an abstract block set for
copying or backup, according to one embodiment of the present
invention.
FIG. 14 illustrates one embodiment of a method for forming an
abstract block set.
FIG. 15 illustrates one embodiment of a method for copying or
restoring a logical object from an abstract block set.
FIG. 16 illustrates one example of combining two forms of metadata
for a logical object into a remapping table for restoring the
logical object to a new area of memory.
FIG. 17 illustrates one example of storage of a logical object
across a computer system and formation of an abstract block set
using physical backup segments corresponding to track size,
according to one embodiment of the present invention.
FIG. 18 illustrates an example of one embodiment of metadata for an
abstract block set that has a physical backup segment granularity
larger than the size of a datablock.
FIG. 19 illustrates one embodiment of a method for backing up a
logical object while preventing updates to the logical object
during the backup.
FIG. 20 illustrates one example of a system that includes markers
for physical backup segments, permitting avoidance of updating
information within a logical object during a backup process,
according to one embodiment of the present invention.
FIG. 21 illustrates one embodiment of a method for performing a
differential backup.
FIG. 22 illustrates one embodiment of tracking changes at the
physical level of a system, and converting those changes to logical
information.
FIG. 23 illustrates an example of performing a differential backup
on a logical object, according to one embodiment of the present
invention.
FIG. 24 illustrates one example of forming full and differential
logical backup objects for backup of a logical object, according to
one embodiment of the present invention.
FIG. 25 illustrates one example of combining an abstract block set
and a differential abstract block set into a single full abstract
block set, according to one embodiment of the present
invention.
FIG. 26 illustrates one embodiment of a method for combining
differential abstract block sets with a full abstract block set to
produce a new full abstract block set.
FIG. 27 illustrates one example of a system for backing up data on
a primary storage node, using a secondary storage node, according
to one embodiment of the present invention.
FIG. 28 illustrates one embodiment of a state diagram for any
synchronous transfer of data for copying or backup.
FIG. 29 illustrates one embodiment of a state diagram for
asynchronous restore of a backed up logical object.
FIG. 30 illustrates one embodiment of a system and data flow within
a system for sending copy of backup information from a primary
storage node.
FIG. 31 illustrates one embodiment of a method for sending data
from a primary storage node.
FIG. 32 illustrates one embodiment of a structure and data flow for
control of writing information to a backup media.
FIG. 33 illustrates one example of a tape media written with backup
abstract block sets.
DETAILED DESCRIPTION
The architectures illustrated in FIG. 5 and FIG. 6 may be viewed as
focusing on a network model for storage, or a "network-centric"
system. In such a system, the focus of data transfer is movement of
logical data across a network. Moreover, the storage system 52 and
backup storage system 54 are typically associated with a single
client or host 50 architecture.
An alternative model focuses on a separation of the client or host
domain and the storage domain.
FIG. 8 illustrates one example of a system which segregates the
host domain from the storage domain. In FIG. 8, a number of host
computers 80 are included in an enterprise host domain 80a. The
host computers can be any type of computers, operating systems and
data management applications. For example, one host computer 80 may
be a Hewlett Packard 9000 computer system running an HP-UX
Operating System. Another host computer 80 can be a Sun Spark
Station running a Solaris operating system. The combination of a
host, operating system and applicable data management application
is referred to as a "platform." Each of the host computers 80 may
constitute a different platform interfacing with the storage
network 89.
The host computers 80 in the enterprise host domain 88 may be
connected over a network. This network may include switching nodes
81, although any other form of network may be used.
In the embodiment of FIG. 8, the host computers 80 are coupled to
the enterprise storage 89 through a network or directly to primary
storage nodes 82. A primary storage node is a memory device that
can store significant amount of data for use by the host 80. For
example, a Symmetrix system, such as the one described above with
respect to FIG. 7, may be used as a primary storage node, although
this is not intended as limiting.
In the embodiment of FIG. 8, each host computer is coupled to a
subset of primary storage nodes 82, for use as a main memory for
that host computer. For example, host computer 80a is coupled
directly to primary storage node 82a. The host computer 80a may
rely on primary storage node 82a for most of its memory intensive
functions, such as for accessing a very large database.
The primary storage nodes 82 may also be coupled together through a
network. In the example of FIG. 8, the network includes link 85 and
switch network 84. The switch network 84 may, for example, be a
fiber channel network. The link 85 may be an RDF link over an ESCON
line.
The network between primary storage nodes may serve two purposes.
The network may permit transfer of data between primary storage
nodes. For example, a database being manipulated by host 80a, and
stored in primary storage node 82a, may be transmitted to primary
storage node 82b for use by host 80b. By transmitting the database
across the enterprise storage network (using link 85 or switching
network 84), the computational resources of the host 80a, 80b, and
the available bandwidth in the enterprise host domain network, can
be preserved.
The enterprise storage network 89 may also include a secondary
storage node 87. The secondary storage node may be used for backup
functions, hierarchical storage management, virtual disks and other
functions. Thus, the secondary storage node 87 may be coupled to a
tape storage unit 83. The secondary storage node 87 would
coordinate sophisticated transfer of data from the primary storage
nodes 82 to the tapes stored in a tape storage unit 83. (Other
embodiments may use additional or alternative media for secondary
storage.)
FIG. 9 illustrates one embodiment of a computer network constructed
according to one aspect of one embodiment of the present invention.
In this embodiment, an enterprise host domain 97 is provided. The
enterprise host domain 97 includes a variety of host computers
90a-90e. The host computers may include different platforms and
different corresponding mechanisms for accessing and storing data
in the memory. For example, host computer 90a is a Hewlett Packard
HP 9000 computer. Host computer 90c is a Sun Spark Station which
may be running a Solaris Operating System. The host computers may
communicate with each other across a network 96. Such a network can
be one of many computer networks known and applied for connecting
computers.
In the embodiment of FIG. 9, each host computer 90a-90e is
connected to a primary storage node 92a-92c. In this embodiment,
each primary storage node 92a-92c is an iterative cached disk
array, such as a Symmetrix memory system such as the one described
above with respect to FIG. 7, although this is not intended to be
limiting. Thus, for example, host computer 90a interfaces primarily
with storage node 92a. Similarly, host computer 90b uses primary
storage node 92a as a primary source of its data.
In the embodiment of FIG. 9, the host computer 90a is connected to
the primary storage node 92a over a high speed fiber channel 91a.
The host 90b, however, is connected to the primary storage node 92a
over a standard SCSI connection. Each of the hosts 90a and 90b are
coupled to the same primary storage node 92a. Other mechanisms
could be used to connect the host computers 90a-90e to the primary
storage nodes 92a-92c. For example, a complete switched network
could be employed, for any of the host computers to access any of
the primary storage nodes 92a-92c.
Each of the primary storage nodes 92a-92c may also be coupled
together using a network. In the example of FIG. 9, the only link
among the primary storage nodes is an ESCON remote data facility
(ESCON "RDF") link 93g. Such a link may be used for transferring of
data or maintaining a mirror of data either on-line or as a
periodically updated mirror. Such a link may be implemented as
described in U.S. Pat. No. 5,544,347 (Yanai), which is incorporated
herein by reference in its entirety. Each of the primary storage
nodes 92a-92c may be coupled together using any other mechanism.
For example, an RDF link could be used to fully connect each of the
primary storage nodes 92a-92c. In the alternative, a switch network
could be used, assuming that the network is of sufficiently high
speed to support the data operations among the primary storage
nodes 92a-92c.
The storage network 98 in the embodiment of FIG. 9 further includes
a secondary storage node 94. The secondary storage node is used for
backup (and other) functions, for example by storing and restoring
information to and from a tape library 95.
In the embodiment of FIG. 9, each of the primary storage nodes is
connected or connectable (by a network) to the secondary storage
node 94. In this example, primary storage nodes 92b and 92c are
coupled to secondary storage node 94 each using an RDF link (93c
and 93d respectively) which may be implemented as described
above.
The primary storage node 92a is connected (together with other
primary storage nodes, not shown) to the secondary storage node 94
over a switched network, which will permit each of the systems to
access the secondary storage node 94.
Using an RDF (or other) link that permits high speed transfer of
data over long distances, the primary storage nodes 92a-92c and the
secondary storage device 94 may be physically located at great
distances apart.
Of course, other topologies and other mechanisms may be used
without departing from the scope of the invention.
Many of the applications for computers now focuses as much or more
on memory than on the ability of the system to perform
computations. For example, access to very large databases has
become an extremely important and valuable application for
computers.
In the past, the focus of computer systems has been on
interconnecting host computers each having their own associated
memory, or providing network access to a single memory. This focus
demands host computer and network resources.
In the storage-centric model, however, the storage component of the
computer system is elevated to a status of equal importance. In
such a model, the storage components of the system are capable
interacting with each other with less involvement from the host
domain. For example, it may be desirable to permit mirroring across
one or more primary storage nodes. Similarly, data objects may need
to be copied from one primary storage node to another primary
storage node. Where additional levels of backup are desirable, the
primary storage nodes may also transfer data to a secondary storage
node for backup purposes. The primary storage nodes may,
correspondingly receive data from the secondary storage nodes for
restore. In a storage centric model, some or all of the resource
intensive functions in such a system can be moved out of the host
domain. Certain embodiments following this model can preserve host
domain resources, increase scalability of memory (by adding to the
storage domain without as much concern about affect on host domain
resources) and reduce dependence on the particular platforms of the
hosts in the host domain.
FIG. 10 illustrates, at a very basic level, how data is moved in
one such system. At a step 100, the physical elements (e.g., data
blocks) that need to be copied, backed up or restored are
identified. At a step 102, those physical elements are
transferred.
For example, for a copy, the physical elements that are to be
copied are identified at step 100. In addition, the location of
where the elements are to be copied to are identified. For a copy
between primary storage nodes, this may involve identifying the
copy from locations and the copied to locations. For a backup, this
involves identifying the copy from locations and may be as simple
as determining what tape or other backup storage element will
receive the backup data.
For a copy between primary storage nodes, the physical elements are
transferred from the identified copy from locations to the
identified copy to locations. For a backup, the physical elements
are copied to tapes. (Although reference is made to tapes as
secondary storage, this is not intended to be limiting. Any other
storage media may be used).
The step 100 can, however, be extremely complicated. In many cases,
it is not desirable to copy the entire contents of a primary
storage node. Rather, only a subset of the physical elements in the
primary storage node may need to be copied. As one example,
consider backing up a database stored in primary storage node 92a
of FIG. 9. This database may occupy only a small portion of the
total data stored in the primary storage device 92a--in fact, there
may be an extremely large segment of data accessible primarily by
the host computer 90b which host 90a may not even be capable of
reading (because it is a different platform than the host computer
90a).
In short, it may be desirable to backup a logical object stored
within a primary storage node. In this case, the step 100 requires
mapping the logical object onto the physical elements in the
primary storage node 92a in order to identify the physical elements
that need to be copied from 92a. As described above with reference
to FIG. 2C, these physical elements may be located in disparate
locations within the primary storage device.
The step 102 may similarly be complicated. Even after all of the
physical elements in the primary storage device have been
identified, simply transferring the physical elements is
insufficient. The relationship between the physical elements may
need to be preserved for the copied or backed-up logical object to
be read by the host computer coupled to the receiving primary
storage node. One mechanism for use of mapping a logical object to
physical elements and preserving the logical relationship between
those physical elements is discussed below. This is not intended as
limiting with respect to other aspects of the present
invention.
In any event, under a storage-centric model of computer storage, it
may be desirable to permit as much of the data transfer process
(e.g., the one shown in FIG. 10) to be performed within the storage
network--and without requiring resources from the host domain.
Accordingly, the primary storage nodes and the secondary storage
nodes in the network may include sufficient intelligence to handle
aspects of the data transfer process. For example, the primary
storage nodes may be capable, at a minimum, of managing the
transfer of identified physical elements in a logical object even
when those physical elements are stored in disparate locations
within the primary storage device. In a storage centric model of a
computer system, it may be desirable to move some (or as much as
possible, in some cases) of the data transfer functions to be
performed using resources among primary and secondary storage nodes
within the storage domain.
The computer system may include a storage management application
("SMAPP") for managing manipulation of storage within the storage
domain. The SMAPP can be implemented using software on the host
computers, primary storage nodes, a separate storage controller or
in some combination of these, as described below with reference to
FIGS. 11A and B, below.
The storage management application can be implemented using three
primary components--a management component, server component and
client component.
The management component controls configuration of the backup,
control and monitoring of the backup and copying processes in the
storage domain. The management component also tracks location of
copies of logical objects in the storage system including, for
example, what tape or tapes contain backups of each particular
logical object.
The server component controls the hardware functions of the memory
process, such as acts of mounting and dismounting tapes, opening
and closing, reading and writing tapes and other memory media.
The client component of the SMAPP handles manipulation and
identification of the backup or copy-from source. For example, the
client component is responsible for identifying the applicable
logical object (e.g., file system, file or database) and
determining what operating system level (or logical volume manager
level) physical elements are involved. (As described above, an
additional layer of mapping may be performed within the storage
domain at the primary storage element of 111. For example, if the
primary storage element 111 is a Symmetrix product as described
above, the identified physical tracks may be re-mapped within the
primary storage element 111.)
FIG. 11A illustrates one example of a portion of a computer system
having a host domain and a storage domain. In the example, only one
host 110 is shown in the host domain. In addition, only three
components are shown in the storage domain. These are the primary
storage element 111 (which may be, for example, a Symmetrix disk
array), a secondary storage element 112 and a tape library unit
113. As described above, additional storage elements may be
included, coupled together by a network. For simplicity, the
example of FIG. 11A shows only one element from each of three
different storage levels--host, primary storage element and
secondary storage element.
In the example of FIG. 11A, a storage management application
("SMAPP") 114 is primarily a resident on the host computer 110.
Thus, the host computer would include an Application Programming
Interface ("API") which would permit management of copying, backup
and restore (and other) operations. In addition, the storage
management application 114 on the host 110 includes a server
component 115b. Again, the host would include an API permitting
management of server operations. Finally, the storage management
application 114, in this example, includes a client component 115c.
The client component would be responsible for identifying and
manipulating logical objects and identifying (from the operating
system or logical volume management level view of) the physical
elements that comprise the logical object.
For simplicity, the operation of performing a backup from the
primary storage element 111 to the secondary storage element 112
will be described. A similar process would apply for setting up
mirroring or copying functions between primary storage elements in
a network.
In this example, the primary storage element includes an SMAPP
interface 116a. Similarly, the secondary storage element 112
includes an SMAPP interface 116b. The copying of a logical object
from the primary storage element 111 to the secondary storage
element 112 in the embodiment shown in FIG. 11A may proceed as
follows. First, a "virtual circuit" or "connection" is set up
between the primary storage element 111 and the secondary storage
element 112. This may be a virtual circuit established through a
network coupling the primary storage element to the secondary
storage element 112 (including a single RDF link between the
primary storage element 111 and the secondary storage 112, for
example). In addition to establishing a physical connection between
the nodes, the virtual circuit identifies a session for copying a
series of data (comprising, e.g., the logical object) over the
identified connection.
Thus, the management component 115a on the SMAPP 114 on the host
computer 110 may begin a backup session by instructing the primary
storage element to establish a virtual circuit with the secondary
storage element 112. The actual establishment of the virtual
circuit may then be performed by the SMAPP interface 116a of the
primary storage element 111 in combination with the SMAPP interface
116b of the secondary storage element 112.
The client component 115c of the host computer 110 identifies a
logical object for backup. The client component 115c then maps that
logical object to the operating system (or a logical volume manager
level) set of physical elements. This mapping may be performed in
one step. The client component 115c of the host 110 may then
identify the elements for copying to the primary storage element
111, as communicated through the SMAPP interface 116a.
The server component 115b of the host 110 would identify and mount
the appropriate tapes in the tape library unit 113. In this
particular example, the server component 115b performs these
commands by passing them to the SMAPP interface 116b of the
secondary storage element 112, through the SMAPP interface 116a of
the primary storage element 111, which then mounts the tapes.
The actual performance of the backup process may proceed, without
further control by the host 110 of the host domain (except, in some
embodiments, monitoring the process and managing the backup media,
e.g., controlling changing of tapes in a tape drive). The primary
storage element 111 may copy the identified physical segments to
the secondary storage element 112.
FIG. 11B illustrates an alternative structure for control of the
storage domain of a computer system according to the present
invention. In this example, a storage network controller 118a is
connected to the host 110, primary storage element 111 and
secondary storage element 112 through a network 119. This network,
for example, may follow the TCP/IP protocol. The storage network
controller 118a may be any hardware, or hardware and software,
combination capable of performing the requisite functions. For
example, the storage network controller 118a may be a computer
running a windows NT operating system, with suitable application
software for performing the SMAPP functions.
In this example, a significant portion of the SMAPP software is
resident on the storage network controller 118a. Thus, the SMAPP
118b of the storage network controller 118a includes a management
component and a server component. Thus, management of the hardware
and media can be performed by the storage network controller 118a,
independent of the host computer 110.
In this example, the host 110 includes an SMAPP 117 to perform
client functions. Thus, logical to physical mapping is still
performed in the host domain by the host computer 110. As the
client component of the SMAPP 117 is responsible for identifying
logical objects and performing logical to physical mapping, this
can be a sensible arrangement. The logical to physical mapping
depends on the particular host platform and the host necessarily
has elements capable of performing the requisite mapping.
In other embodiments, however, the client component can be included
in the storage network controller 118a, or in a separate device
capable of performing logical to physical mapping for one or more
platforms. Where this is done, the identification and transfer of
data for copying and backup purposes can be performed completely
separately from the host domain. In many systems, however, it will
be more efficient to use the memory mapping mechanisms (client
component) on the host computer.
Other arrangements of the SMAPP software are possible. For example,
the components of the SMAPP software may be distributed across the
primary storage elements in the storage domain, the secondary
storage element or elements in the host domain or some combination
thereof.
FIG. 12 illustrates one embodiment of a method for transferring a
logical object according to a system such as the one shown in FIGS.
11A and 11B. At a step 120, a virtual circuit is established. As
described above, this may correspond to establishing a physical
connection between the element being copied from (e.g., a primary
storage element) to the storage element being copied to (e.g., a
secondary storage element). In addition, this step 120 corresponds
to establishing a session for performing the copying over the
connection. As described above, the establishment and managing of
the virtual circuit can be performed by an SMAPP component resident
on a host computer, storage network controller, or other
device.
At a step 121, the logical object is mapped to identify the
physical elements being copied from. For performing a backup, this
would correspond to mapping an identified logical object at the
application level to a set of physical elements at the storage
level.
To restore from a tape, this would correspond to identifying the
logical locations of the segments of memory on the tape. If the
tape contains a logical bit file, this step is straightforward. No
actual mapping needs to take place. In other circumstances, such as
the abstract block sets described below, a table or other structure
may identify the mapping of portions of the physical elements to
their order in the logical object. The actual mapping from the
logical level to the physical level may have been performed at the
time of the backup and saved.
At a step 122, update to physical elements is prevented. For
example, if a database is being backed up from a primary storage
element to tape, updates of the logical object should be prevented
so that the backup can correspond to a single point in time. Of
course, if the copying is from a backup tape to a primary storage
element, the freezing of updating the physical elements is rather
simple--the tape will not be written while it is being read from in
the restore. In one embodiment, a method for concurrent copying
described below may be used to prevent the update of physical
elements during the copying process.
At a step 123, the copy-to memory is managed. For a backup from a
primary storage element to tape, this may correspond to mounting
and disbounding the appropriate tapes, as well as managing the tape
library, catalog information, as well as writing appropriate tape
headers and trailers. Where the information is being copied to
another primary storage element, this may correspond to managing
the receiving physical elements of the primary storage element
being copied to. In addition, it may involve setting up an
appropriate storage area to receive the information.
At a step 124, the actual physical elements are copied. The copying
may be done in the appropriate order for the logical object, such
as when an ordinary data file is sent at the application level
between two host computers. In the context of a backup, one such
system is described in U.S. patent application Ser. No. 09/107,679,
which is incorporated herein in its entirety. In an alternative
embodiment, the physical data blocks may be copied out of order,
together with appropriate metadata identifying the correct order of
the physical elements in the logical object. An embodiment of this
type of system is described below.
At a step 125, the physical elements of the logical object, in the
copy-from memory, are unfrozen--allowing updates of the logical
object. The backup is complete and the physical elements can be
unfrozen.
Finally, at a step 126, the virtual circuit may be closed.
Logical Object Translation to Abstract Block Sets
As described above, there are at least two different ways of
passing data blocks of a logical object to a storage
element--transferring the blocks in order as a logical object (as
is done over a network between host computers) and a pure physical
copy (which may not preserve the logical relationship among the
data). Each of these possibilities has advantages and
disadvantages. For example, copying each data block of a logical
object in order preserves the relationship between data blocks. On
the other hand, copying the blocks in order may result in delays as
the storage elements sequentially retrieve the data blocks or sort
the data blocks for writing, as a part of the copy process. On the
other hand, pure copying of physical elements can be unnecessarily
slow if unused physical elements are copied. In addition, the
logical relationship between the data blocks that are copied may be
lost.
An alternative is to use an abstract block set structure, as
described more fully below. This type of structure is useful not
only in the storage network architecture as described above, but
has greater applicability. For example, the abstract block set
concept may be employed in any system where logical objects are
copied from one storage element to another storage element. The
abstract block set can also be used to particular advantage when
used for backing up and restoring data from a secondary storage
device, such as a tape drive.
The abstract block set permits storage of the data blocks in any
order. The abstract block set includes information about the
ordering of those elements.
FIG. 13 illustrates one example of an abstract block set. From the
application perspective, a logical object 130 includes a number of
data blocks 130a-130f (ordinarily a logical object may include
substantially more data blocks, FIG. 13 being by way of
illustration only). The data blocks having a logical relationship
or order, as illustrated by labels A-F in the logical object
130.
The logical object is stored in a physical memory 131, as generally
described above with reference to FIG. 2C. Each column may be
viewed as a track (although this is not intended as limiting), and
each row as a row of blocks within the tracks. As shown in FIGS. 2C
and 13, the logical data blocks may be scattered throughout the
physical memory 131.
An abstract block set 132 may be constructed from the data blocks
130a-130f. In the abstract block set 132, the data blocks are not
necessarily stored in the same order as they appear in the logical
object. In this example, they are in a random or pseudo-random
order. (As a practical matter, the order of data blocks may reflect
the way that the data blocks are stored in a physical storage 131.
For example, if data blocks A and B are stored on one track they
would probably be read and written to abstract block set 132 in the
order they appear on that same track. The abstract block set 132
appearing in FIG. 13 is for illustration only.)
Because the logical data blocks are not in order in the abstract
block set 132, it may not be possible to reconstruct the logical
object given only the data blocks 132a-132f.
Accordingly, the abstract block set 132 includes metadata 133. The
metadata is any recorded information that provides a mechanism to
reconstruct the order of logical data blocks as they appear in the
logical object 130.
In the example of FIG. 113, the metadata 133 includes an ordering
of logical block elements (the column labeled LBEL) with the
physical element location. Thus, logical block element 1 has
metadata corresponding to the address of that logical data block in
the physical memory 131--the physical element address. Using the
metadata illustrated at 133, each of the stored data blocks
132a-132f in the stored abstract block set 132 would need to
include a label with the corresponding physical address. Thus, for
example, to locate the first logical data block 130a of the logical
object 130, one could examine the metadata 133 and determine that
the first abstract block set (as shown in the first column of the
metadata 133) has a physical address ADDR-A. This data block could
then be found in the abstract block set 132 by examining the
physical addresses of the data blocks 132a-132f (the physical
addresses appearing within the data blocks 132a-f), until the
appropriate block is found.
Of course, there are a variety of other formats that could be used
for the metadata. As one example, a label other than the physical
address could be used. As another the metadata 133 could just
describe the order of the logical block elements in the abstract
block set 132. In this case, the second column of the first row of
the metadata 133 could indicate that the first logical data block
(corresponding to A) is stored as the sixth block in the abstract
block set 132.
For each of these alternatives, the first column of the metadata
133 is not required. The order of the elements in the second column
corresponds to their location within the logical object 130; the
address for the first logical block element appears first in the
table, the address for the second logical data block appears as the
second entry in the second column, etc.
Metadata 134 illustrates another way of storing the metadata
associated with the logical block 132. In this table of metadata, a
first column corresponds to the ordering of data blocks as they
appear in the abstract block set (as above, unnecessary as the
order that the rows appear implies this information--the first row
is the first block in the abstract block set). The second column
indicates the position of the data block within the logical object
130. Thus, the first entry in the first row of the metadata 134
corresponds to the data block 132a of the abstract block set 132.
This is the second data block 130b of the logical object 130.
Accordingly, the second column has a "2" indicating that this data
block 132a is the second data block of the logical object 130. The
last column of the metadata 134 provides the physical address for
the applicable data block in the physical memory 131.
Using the metadata shown at 134, there would be no need to store
the physical address of the data block with (or other tag) with the
data blocks as stored with the abstract block set 132.
As above, using the metadata 134, it is not strictly necessary to
store the physical address within physical memory 131 of the
applicable data block. This may, however, be useful information to
include within the metadata 134. In many cases, restores will be
made to the same memory locations from which the information was
backed up. In this case, it will be easier to restore to those
addresses in the physical memory 131--that information was not
available. Otherwise, a logical to physical mapping step may be
required to determine again where the appropriate addresses are for
the restored data blocks.
Other formats of metadata may be used. For example, metadata may be
tracked for extents (sequences of blocks) rather than individual
blocks.
FIG. 14 illustrates one embodiment of a method for copying a
logical object to form an abstract block set as described above. At
a step 140, the logical object is identified. As described above,
the logical object can be any logical entity, such as a database, a
segment of a database, file, or file system.
At a step 141, the logical block elements or logical data blocks of
the logical object are identified. This may precede as generally
described above.
At step 142, the logical block elements are mapped to physical
backup segments. The physical backup segments may correspond to the
physical elements that store the logical data blocks. In the event
that the abstract block set is to include metadata of the form
illustrated at table 133, the mapping step 142 may include
formulating that information into whatever format the metadata is
stored in.
As described above, the steps 140-142 may be performed by a client
component of a storage management application. In some systems,
this may require the resources of a host computer.
The remainder of the copying process may proceed without
significant involvement of the client component of the storage
management application.
At a step 144, is to determine whether all physical backup segments
have been copied. If so, the copying process is complete at step
145.
If not all of the physical backup segments have been copied, the
next available backup segment is copied at step 146. As described
above, this copying need not be performed in the order appearing in
the logical object identified at step 140.
In the event that the metadata is being stored as shown at table
134 of FIG. 13, then the metadata may be updated after the
applicable backup segment has been copied into the medium holding
the abstract block set. For this form of metadata (but not the form
shown at 133 of FIG. 13). This may not occur until the applicable
backup segment is copied to the medium storing the abstract block
set because, until that time, the order of appearance for the
applicable physical backup segment is not known.
FIG. 15 illustrates one embodiment of a method for restoring an
abstract block set to a memory system, such as the primary storage
node described above.
At a step 150, the metadata for the abstract block set is
retrieved. This may be in the form of a map for the abstract block
set such as those illustrated at 134 of FIG. 13 or may be a set of
labels associated with the individual data blocks stored in the
abstract block set, such as in table 133 of FIG. 13.
At a step 151, memory is allocated in the target storage device for
receiving the logical object. The amount and configuration of the
memory required to receive the logical object can be determined
from the metadata for the abstract block set. Thus, the metadata
will include sufficient information to determine the
characteristics of storage required. For example, in the event that
the abstract block set indicates use of fixed size blocks, the
total number of (fixed size) blocks required to store the logical
object can be determined by the number of entries and a metadata
table or maybe separately stored as a part of the metadata for the
abstract block set.
At a step 152, dummy metadata is created for the newly allocated
physical memory for the logical object to be restored. The result
can be a new table such as the one shown at 133 of FIG. 13.
At a step 153, a re-mapping table is created. The re-mapping table
specifies a correspondence between the data blocks of the abstract
block set is the source of data and the allocated data blocks in
the physical memory. An example of a re-mapping table is described
with reference to FIG. 16. Although shown in tabular form, the data
can be stored in other forms and formats.
At a step 154, it is determined whether all of the physical backup
segments have been restored from. If so, the restore is complete at
a step 155.
If not, at a step 156, the next physical backup segment is
retrieved. At a step 157, the location and the newly allocated
memory for receiving the logical object is determined. This can be
done by examining the re-mapping table created at step 153. In
addition, the retrieval of segments done at step 156 need not be in
any specific order. The re-mapping table permits restoration of the
entire logical object even when the data blocks are provided in a
random order.
At a step 158, the data from the physical backup segment is
restored to the appropriate locations. Steps 154-158 then continue
until all of the data blocks have been properly restored.
FIG. 16 illustrates an example of creation of a re-mapping table.
Of course, many variations on the creation of re-mapping table are
possible, depending on how the metadata is formulated and stored
for the abstract block sets.
In FIG. 16, metadata 160 is provided for the abstract block set
that is serving as the source for the restore. This table
corresponds to the metadata 133 of FIG. 13.
FIG. 16 also illustrates dummy metadata 161 for the allocated
memory that will receive the restored logical blocks of the
restored logical object. In this embodiment, the format is the same
as that for the metadata 160, except that different addresses
(potentially on a completely different storage element) are
provided. Thus, for the first row in metadata 161, the first
logical data block should be stored at the physical location
specified at ADDR-AA.
A simple merging of these two tables can result in a re-mapping
table 162. The re-mapping table 162 specifies the physical location
from the data in the abstract block set and the destination for the
that logical data block.
Of course, other formats may result in other tables. For example,
it would be possible not to specify any physical addresses in the
re-mapping table 162. The re-mapping table could simply map the
sequential location in the abstract block set being restored from
to the physical address or to the sequential location on the
receiving storage element.
In other embodiments, each entry in the metadata remapping table
may correspond to extents in the physical memories restored from
and to.
Physical Backup Segment Granularity
In the discussion with respect to FIGS. 13-16, it was assumed that
the backup, copy and restore was performed at the data block level.
Thus, the physical backup segment corresponded in size to the size
of a data block. Those data blocks that are part of the logical
object, and only those data blocks were copied for backup and were
restored.
Granularity of the physical backup segments need not, however,
correspond to the granularity of the data blocks. For example, a
track may store a number of physical data blocks. In some
instances, not all of the data blocks within a track are
necessarily a part of the same logical object. Thus, in a track
that stores four data blocks, only two of those data blocks maybe a
part of a logical object, the other two data blocks being unused or
part of a different logical object. Backing up of data in a logical
object may, however, be performed at the track level rather than
the physical data block level. The result would be an abstract
block set that includes some data blocks that are not a part of the
logical object.
Thus, in the preceding example, the physical backup segment size
corresponds to the size of a track. The actual physical data blocks
that may store the data of a logical object are smaller, e.g, four
data blocks per physical backup segment of one track.
FIG. 17 illustrates the concept of a physical segment size (here, a
track) that is larger than the size of the physical data blocks. In
the example of FIG. 17, a logical object 170 is stored on a
physical device that includes tracks. Each track holds (in this
example) up to three data blocks.
At the application level, the logical object 170 is viewed as a
continuous file. This file may be partitioned into a number of
logical data blocks, shown in FIG. 17 as vertical bars within the
logical object 170.
At the file system level, a file system image 171 holds that data
in each of the logical data blocks of 170. As shown in the file
system image 171, the order of the logical data blocks at the file
system level may not correspond to the order of their appearance
within the logical object 170. As described above, a mapping
process maps the logical data blocks to appropriate locations
within the file system image 171.
The file system image 171 may be mapped to a logical volume of
hypervolume level 172a-172b.
The logical volumes 172a-b are then stored on a physical storage
device in hypervolumes 173n and 173o. As shown in FIG. 17, the
hypervolumes may not be physically adjacent. (Of course, as
described above, other techniques for mapping the logical data
blocks of the logical object 170 to the physical storage device are
possible and within the scope of the present inventions.)
The first hypervolume 173n stores data across seven tracks
173a-173g. These tracks may, but need not, be contiguous segments
of memory.
In this example, the entire track 173b contains physical data
blocks that are part of the logical object 170 (given the
assumption that only three data blocks are stored per track). The
track 173d, however, includes only one data block that is a part of
the logical object 170--the other data blocks in the track 173d
either being unused or containing data belonging to a different
logical object. In addition, some of the tracks within the
hypervolume 173n do not contain any data from logical object 170,
e.g., tracks 173a, 173c and 173f. The hypervolume 173o similarly
contains some tracks that include data from the logical object and
some tracks that do not.
Given that the physical backup segment granularity is chosen to be
track size in this example, the physical segments that would be
part of a backup process would include tracks 173b, 173d, 173e,
173g, 173i, and 173k. These tracks make up the physical backup
segment set (here, a "trackset") that would be copied when the
logical object is backed up. Since, in the example of FIG. 17, the
physical backup segment granularity is by tracks, this may be
referred to as a track set.
Thus, the track set for a backup of logical object 170 would
include tracks 174a-174g, which in turn correspond to those of the
physical tracks 173a-173m that include data blocks from the logical
object 170.
The backup process using a physical backup segment size that is
different than the data block size can proceed generally as
described with reference to FIG. 14. At step 142, however,
identification of the physical backup segments includes not just
identifying the logical block elements but using the identified
logical block elements and their physical data block locations to
determine the physical backup segment set, e.g., the track set
174a-174g of FIG. 17.
In addition, the copying of the available backup segments at step
146 would involve copying the larger granularity segment (e.g., a
complete track rather than just the particular physical data blocks
on the track). As in FIG. 14, the physical backup segments (e.g.,
tracks) may be copied in any order.
Returning to FIG. 17, an abstract block set signature track 175 may
be stored. This signature track includes the metadata for the
abstract block set. In this embodiment, specification of the
metadata for the abstract block set may include a specification of
the particular data blocks in the abstract block set and their
location within the logical object 170.
FIG. 18 shows one example of metadata 180 for an abstract block set
that has a physical granularity greater than the size of the
physical data block. In this example, the location of each data
block is specified. The first column is a specification of the data
block within the logical object--e.g., first, second, third, fourth
data block.
The second column of the metadata 180 specifies the physical
address of that logical object. In this example, that physical
address includes a specification of where within the physical
backup segment the applicable data block is located. For example,
this information may be included as an offset within the physical
backup segment. Thus, an address of d:cccc:h:offset includes a
specification of the physical backup segment (dd:cccc:h), which in
this example specifies a track and a location within that physical
backup segment (track), and an offset. For example, the first row
of metadata 180 corresponds to the first logical data block in the
logical object. It also happens to appear as the first data block
in the specified physical backup segment address, e.g., as an
offset from the beginning of the physical backup segment (here, a
track) of just zero. The second row of the metadata 180 specifies
the same address, but has an offset of 1--it is a data block
appearing in that physical backup segment (track) immediately
following the data block corresponding to the first logical data
block of the logical object.
In this example, it may be assumed that the track that includes the
first two logical data blocks (first two rows of metadata 180) has
additional room within the track, but that those additional data
blocks in the track are not a part of the logical object.
Accordingly, there is no entry in the metadata table 180 specifying
a corresponding logical data block for that portion of the track.
(In an alternative embodiment, of course, an entry could be made
which indicates that that portion of the track is unused in this
abstract block set.)
As described above with reference to FIG. 13, many other forms and
formats for storing metadata may be applied.
Restoring (or copying) from an abstract block set that has a
physical backup segment granularity larger than the data block size
may proceed as generally prescribed above with reference to FIG.
15. At step 157, however, the locations of the logical data blocks
within the backup segment are identified--including determining
whether any portions of that backup segment may be omitted. At step
158, only those data blocks that are actually used in the logical
object are restored.
Selecting a physical backup granularity larger than the size of
data block can require transfer of more memory than if the physical
backup segment size is the same as the data block--some unused data
blocks are included as a part of the copy or backup process.
A larger physical granularity size can, however, result in certain
advantages. For example, less overhead may be required in the
copying process--fewer segments for copying need to be specified.
In addition, other resources may be preserved. For example, if high
demand memory is used for storing information identifying the
physical backup segments to be copied, less such memory is
required. In the event that the physical data blocks of logical
objects are prevented form being updated during a copy or backup
process, this can be done by protecting updates at the physical
backup segment level rather than the data block level--again
requiring less overhead. In some systems, this can reduce the
complexity of avoiding updates. Some embodiments of the invention
employing different physical backup granularity than data block
size may achieve some or all of these advantages, depending on the
context and system in which it is implemented. None of these
advantages is intended to limit the scope of the invention, which
is defined by the claims below.
Concurrent Copy or Snapshot Facility
As described above with reference to FIG. 12, certain systems
embodying one or more aspects of the present invention will allow
copying or backup, of a logical object at a specified point in
time. To do this, updates to the logical object need to be
prevented during the copying or backup process. There are many ways
to do this, including taking the application that uses the logical
object off-line until the backup process is complete. While certain
embodiments of the present invention will use this and other
techniques, it may be advantageous to be able to continue
processing during the backup.
FIG. 19 illustrates one embodiment of a method for performing a
backup while preventing updates to those physical data blocks that
are part of the logical object being backed up.
At a step 191, the logical object (or system using the logical
object) is quiesced. There are at least two ways to quiesce the
system. One way is to take the application off-line, and update the
logical object off-line. This prevents any further updates (or
reads) to the applicable logical object. Taking the application
off-line can be undesirable--resulting in loss of time and system
availability.
An alternative way of quiescing a system is to place the
application in on-line backup mode. For example, if the application
is using an oracle database, writes to the database can be stored
in a re-do log, rather than actually writing the data to a physical
storage system. When the application is brought back to on-line
mode, the updates to the logical object that are stored in the
re-do log may then be applied to the current copy of the logical
object.
At a step 192, those physical backup segments that contain data
from the logical object are marked for copying. This may be done in
a number of ways. For example, a bit may be associated with each
potential physical backup segment in the system. The bit may be set
to a "one" if the corresponding physical backup segment is part of
a logical object to be copied. Of course, the methods for
identifying the physical backup segments that are part of the
abstract block set being copied can be used. As just one example, a
list of the physical backup segments could be maintained.
FIG. 20 illustrates an example of a system having bits associated
with physical backup segments. In the example of FIG. 20, the
physical backup segment size is a track of the physical memory 28,
such as column 28b. A logical object 27 is stored across the
physical memory 28. A series of bits 29 is associated with the
tracks of the physical memory 28. A one is set for those tracks
(physical backed up segments) that are part of the track set for
the logical object. Thus, the second bit of the bit set 29 is set
at one, reflecting the fact that track 28b is included in the track
set for the logical object 27. If the physical backup segment
granularity were a data block, a bit could be associated with each
data block--at higher overhead.
Returning to FIG. 19, the application using the logical object can
be returned to the active state. This may, for example, involve the
step of returning the application to on-line mode from off-line
mode. If the application was held in on-line backup mode, any
elements in the re-do log may be applied to updating the logical
object.
An attempt to write to a physical backup segment included in this
set of segments to be backed up will, however, momentarily stall.
Before the write takes place, that segment is copied or backed up,
e.g., according to the illustrative embodiment described below.
At a step 194, it is determined whether there is a hit on a
physical backup segment that is included in the backup segment set
in the abstract block set. If so, that segment is copied out of
turn--and before the update is made. After the segment has been
copied, that segment can be unmarked--further updates may be
allowed for that segment. After the segment has been unmarked, the
update may be performed. Processing will then continue at step 194
in case there are additional hits (attempts to write to) a physical
backup segment included in the abstract block set.
The copying of the segment may occur directly to the target
(receiving primary storage element or receiving secondary storage
element such as a tape) or may be copied to a cache for later
copying to the target destination of the abstract block set.
If there are no pending hits on the physical backup segments of the
logical object that remain to be copied, then processing may then
continue at a step 196. At this step, it is determined whether all
of the physical backup segments have been copied. If so, the
formation of the abstract block set is complete and processing may
conclude at step 197.
If there is additional copying to be done, the next available
physical backup segment may be copied, at a step 198. Where
abstract block sets are used, which permit the physical backup
segments to be included in any order, the selection of the next
segment may focus on whichever segment is next available,
independent of order.
As before, after the segment has been copied, it may be unmarked.
Accordingly, any incoming writes to that segment, which occur after
the segment has been copied, may be performed--even if the backup
process is continuing with other physical backup segments.
In situations where the physical backup segment granularity is
larger than the physical data block size, a write may occur to a
physical backup segment that does not correspond to a write to a
logical object. For example, consider a physical backup segment
that has one physical data block that is in the logical object that
is being backed up and three other physical data blocks that belong
to other logical objects. A write to one of the physical data
blocks corresponding to different logical object would trigger
backup of the physical data segment, even though the logical object
being backed up is not being updated.
One alternative for handling such a circumstance is to examine each
write to a marked physical backup segment to determine whether the
write is to a physical data block that is a part of the logical
object. While this method may be employed in some embodiments of
the present invention, it can incur a heavy overhead penalty in the
event of writes to physical backup segments.
In an alternative embodiment, the physical backups segments are
treated the same whether or not a write occurs to a physical data
block in the logical object being copied or the physical data block
in a different logical object. The overhead associated with this
alternative may not be great, particularly if implemented in a
system where the copied physical backup segments are being stored
in an abstract block set that permits physical backup segments to
be transferred in any order.
In most cases, it will be easiest to mark and prevent updates to
portions of physical memory based on physical backup segment
granularity--e.g., using tracks on a disk for physical backup
segment size and also for marking and preventing premature updates
to the stored logical object. Other alternatives may be
implemented. As just one example, in a disk system, tracks could be
used as physical backup segments, but prevention of updates marked
or tracked at the data block level (rather than the track
level).
Differential Backups
Systems similar to FIG. 5 and FIG. 6 conventionally backup an
entire logical construct or element ("logical object") specified by
the user. For example, a user may specify a partition to be backed
up. When this is done, the entire partition is copied to the backup
storage system. Similarly, the user may specify a database or file
to be backed up. In this case, the entire database or file is
copied to the backup storage system.
This can require a significant amount of time, which is ultimately
unnecessary. For example, an extremely large file may be backed up
at one point in time. A second backup may then be performed at a
later time. Very little of the file may have been changed between
the two backups. Generating a new backup of the complete file can,
therefore, be wasteful.
FIG. 21 illustrates one embodiment of the present invention for
creating a differential backup. A differential backup is a backup
of only a portion of a logical object, based on what has been
changed during operation of the computer system.
At a step 470, a level zero backup is performed. A level zero
backup is a complete backup of the logical construct. For example,
a level zero backup of a file backs up the entire file. A level
zero backup of a (virtual) partition backs up this entire
partition. Thus, a level zero backup can be used to restore the
logical object without any further information.
At a step 472, the system tracks changes in data from the last
level zero backup. For example, referring to FIG. 2B, the segments
that included any changed data may be tracked. If segments 1, 3 and
5 include data that was changed, a corresponding bit or other
record could be set indicating that these segments have changed
(and not the others). As described more fully below, the segments
may be defined by how the data is physically stored (e.g., by
storage blocks) rather than based on logical level information, and
may (but need not) correspond to the granularity of physical
back-up segment of abstract block sets or the granularity of
physical segments marked to prevent updates.
At a step 474, those data segments that have been changed are
backed up. By backing up only changed data segments, rather than
the entire file, the generation of the backup may be performed much
more quickly. One embodiment of a method for storing and recovering
files using records of just changed data segments is discussed
below. This backup may be referred to as a "differential backup"
because less than all of the logical data blocks are backed up,
e.g., some data segments that have not been changed are not backed
up.
At a step 476, it is determined whether a new level zero backup
should be generated. If not, the system continues tracking changes
from the last level zero backup, at step 472. In this embodiment,
therefore, the differential backup generated at step 474 always
records changed data from the last level zero backup--not from the
last differential backup. An alternative embodiment is to track
changes from the last differential backup.
If a new level zero backup is to be generated, at a step 478, the
tracking of changed data is reset. This may be performed, for
example, by resetting "change bits" associated with the data
segments, described below. While this is done, the system may be
taken off-line or placed in backup mode to assure that data is not
changed while the change bits are being reset (and the level zero
backup performed). When a new level zero backup is performed,
future changes will be tracked from that level zero backup rather
than an earlier one.
In another embodiment, resetting tracking of changed data may be
performed after the step 474 of backing up changed data segments.
In this case, each differential backup tracks changes only from the
last differential backup. As discussed below, to restore data when
this is done, multiple differential backup files may need to be
examined, rather than just the one differential backup performed at
the time of interest.
The complete or level 0 backup may be performed as described above.
For example, an abstract block set may be created, using physical
backup segments stored in any order, together with metadata as the
level of backup.
The step 472 may be performed at either the logical or the physical
level. At the logical level, the client 50 may track the segments
that include changed data. At the physical level, the storage
system 54 may track which segments of data have been changed. In
either case, the segments of data may correspond to physical
segments of data that are stored on the storage system, rather than
units of data (e.g., files within a partition) determined at the
logical level and associated with a logical construct.
The physical segment may be a 512 byte block that is written to or
read from the physical storage device at one time. In another
embodiment, the granularity of the physical segment may be the
amount of data stored in a track of the physical storage devices
used (particularly when the physical storage devices are disk
drives). The size of this may depend on the particular format for
storing data in applicable operating system. For example, in a
fixed block architecture environment, the track may be 32 kilobytes
(64 SCSI blocks). On IBM main frames implementing a count-key-data
("CKD") system, the segment size may be the size of one CKD track.
As above, the granularity of the physical segments for which
changes are recorded may, but need not, correspond to the physical
backup segment size or the granularity at which updates are
prevented during the copying or backup process. In many cases,
however, it will be most efficient to use the same granularity for
each of these functions, e.g., using a physical track on a disk for
the granularity of the entire system.
In certain embodiments, the changed segments may be tracked at the
physical storage level. Thus, whenever a physical segment is
written to a physical storage device, the fact that the segment was
changed can be recorded. This may be done using a single bit
associated with each physical segment. When the system is
initiated, all of the bits are set to zero (for example). When a
physical segment is changed (or written), the associated bit may be
set.
Thus, referring again to FIG. 2B, data changes may be tracked at
the level of the actual physical storage devices 204-206. When data
is changed in one of the data segments, a bit may be set (or some
other mechanism used) to track that that segment has been changed.
For example, if data is changed within the first segment of the
application file at the application level, e.g., 203a, the data in
actual physical storage device at 208 will be modified. A bit (or
other mechanism) associated with data segment 208 will be set when
this write is performed.
FIG. 2 illustrates one example of a system that includes a bit
associated with physical storage segments. For example, physical
storage device 201a includes six physical segments.
An associated physical bit mask 412 sets a bit for each physical
segment that has been changed. In this example, segments 114a and
114b have been changed. Accordingly, the associated bits 412a and
412b of the physical bit mask 412 have been set to one. On inquiry,
the physical bit mask may be read and output to a client (e.g.,
client 50 of the system illustrated in FIG. 5).
The actual physical storage devices 204-206 may, but need not, have
any idea of what is being done at the application level. In this
embodiment, the physical storage devices need only be aware that
data within the applicable segment of data (e.g., 208) is being
modified.
(While many of the embodiments described herein use bit masks to
represent changes in data, e.g., a physical bit mask or a logical
bit mask, other mechanisms (lists being just one example) may be
used.)
In the embodiment described above, the changes to data segments are
tracked at the physical storage level (although, in alternative
embodiments, the changes could be tracked at any of the other
levels, e.g., the application level, file system level, logical
volume or logical volume manager level, as illustrated and
discussed with respect to FIG. 1).
In one embodiment of performing a "differential" backup, data about
changes at the physical level is converted to correspond to changes
at the logical (e.g., application file) level. The differential
backup then stores the data at the logical level.
FIG. 22 illustrates one way of tracking changes at the physical
level and converting that to the logical level. In this embodiment,
a bit mask 412, 413 and 414 is associated with each actual storage
device 204-206.
When data is written to a data segment, a corresponding bit and the
corresponding physical bit mask is changed from a zero to a one.
Accordingly, at any point in time, the physical bit masks indicate
all of the data that has been changed since the last backup. As
described above, the actual physical storage devices 204-206 may
not know how this corresponds to logical objects at the application
level. Indeed, the actual physical storage devices may have no way
to determine what data segments are associated with each other. As
indicated in FIG. 22, in this embodiment, data segments 114a-114d
have been changed. Accordingly, corresponding bits 412a, 412b,
412c, 412d in bit masks 412, 113 and 114 have been set to one.
(Other data segments in the actual physical storage devices may
also have been changed, but are not shown in FIG. 11).
A logical bit mask 410 may be constructed, which indicates what
data segments within the application level file have been modified.
Thus, logical bit masks 410 may include entries 410a-410d
indicating that the corresponding data segments 411a-411d have been
altered. (In an alternative embodiment, the segment changes may be
tracked at the logical level, even though the segment size
corresponds to a physical storage amount, such as block or track
size.)
The logical bit mask 410 can be constructed using mapping 202. In
particular, the mapping 202 may convert the application level
object to a group of data blocks in the actual physical storage (as
this needs to be done to store the application level file in
physical storage in the first place). Thus, the mapping 202 may be
performed using the same mechanisms for mapping application level
data into physical storage devices (through, e.g., levels 10, 12,
14 and 16 of FIG. 1). The physical bit masks associated with these
data segments on actual physical storage may then be examined. A
logical bit mask can be constructed by setting each entry in the
logical bit mask to a one only where the actual physical storage
device indicates that that data segment has been changed.
FIG. 23 illustrates one embodiment of the method for performing a
differential backup of an abstract block set. In this embodiment,
the affected memory in the actual physical storage devices is first
quiesced, at a step 231. Quiescing the memory assures that no
additional data is modified within the application level file.
Quiescing may be performed as generally described above, e.g., by
taking the application off-line or placing the application in
on-line backup mode.
At a step 232, a logical to physical mapping is performed to
determine which physical data segments within the physical storage
device are of interest. The step 232 may be performed as generally
described above. That is, using the application, file system and
logical volume manager (where present, and additional levels of
mapping if present) to map all of the data segments within the
application file onto physical storage. As described above, this
may map the object all the way down to actual physical storage. In
other embodiments an additional level of mapping may occur before
reaching the actual physical devices storing data; for example, in
a Symmetrix product as described above, the Symmetrix product may
present what appears to be a three volume storage device. This
Symmetrix product could present change data based on that three
volume set. On the other hand the way the data is actually
physically stored within the Symmetrix may not correspond to that
three volume set provided to the application or operating system
level. Thus, an additional level of mapping for both data segments
and bit masks may be performed within the storage device.)
The granularity at which the changes to data is tracked may be
based on the size of the data blocks or on a different granularity,
such as the size of physical backup segments. For example, change
data may be tracked corresponding to physical tracks, when the
physical data block size is less than an entire track.
At a step 233, the physical data segments that have been changed
since the last time mark are identified. This may be done by
examining the physical bit masks associated with the physical
storage devices. Any entry marking changed data in the physical bit
mask that corresponds to a physical backup segment within the
application that includes a physical data block in the applicable
logical object corresponds to data that may have been changed. At
step 232, a logical bit mask may be constructed, such as the
logical bit mask 410 of FIG. 22.
At a step 234, a differential abstract block set is created. This
step involves copying only those physical backup segments that may
include changed data. In one embodiment, as for the abstract block
sets above, the abstract block set may record the physical backup
segments in any order.
Accordingly, at a step 234, metadata for the differential abstract
block set is also stored. This metadata records information
sufficient to identify the applicable location of the physical data
blocks stored in the differential abstract block set within the
logical object being backed up or copied.
Finally, at a step 236, the application is returned to active mode.
That is, the system is allowed to continue updating the physical
data blocks on the actual physical storage devices.
As described above, before returning the system to active mode, the
bits corresponding to the backed up data segments on the actual
physical storage device may be reset to zero. This is only done if
the differential backups are being performed with respect to the
last differential backup. Otherwise, the bits may only be reset
after the construction of the real (or merged, as described below)
level zero backup.
FIG. 24 illustrates an example of creation of a differential
abstract block set according to the method of FIG. 23. The logical
object 240 includes five physical data blocks. (For simplicity, the
physical backup segment and physical data block size are assumed to
be the same in FIG. 24. As above, however, the physical backup
segment size may be a size that is larger than the physical data
blocks.)
At an earlier point in time, an abstract block set 242 was formed.
As above, the abstract block set stores each of the logical data
blocks of logical object 240, but in any order. The abstract block
set 242 may include metadata, specifying the locations of the data
blocks within the logical object.
After the abstract block set 242 was formed, additional changes may
have been made to the data within the logical object 240. In this
example, logical bit mask 241 reflects those changes. In
particular, logical bit mask 241 indicates that the second and last
logical data blocks within logical object 240 have been
changed.
The differential abstract block set 243 stores those data blocks
that have been changed (the second and the fifth). As described
above, these may be stored in any order. The differential abstract
block set may include metadata for the differential abstract block
set. In the example of FIG. 24, the metadata is of the same general
format as the metadata for the full abstract block set 242. The
metadata includes an extra column, however, that specifies which of
the logical blocks have been changed since the last backup (again,
the second and the fifth, in this example).
To restore a logical object from a full abstract block set backup
and a differential abstract block set, the two may be combined or
merged. In fact, an abstract block set and one or more differential
abstract block sets may be merged at any point in time, off-line.
This permits formation of a synthetic full abstract block set that
reflects the state of the logical object at the point in time when
the differential abstract block set was formed.
FIG. 25 illustrates an example of this merging process, using the
example of FIG. 24. As can be seen, the original data blocks 242a-b
of the whole abstract block set 242 have been updated in the
differential abstract block set 243. Accordingly, in the merged
abstract block set 253, these data blocks have been replaced with
the updated version.
FIG. 26 illustrates one embodiment of a method for performing this
merging process. In the embodiment of FIG. 26, one or more
differential abstract block sets may be present. More than one
differential abstract block set may be present if, for example,
differential abstract block sets are formed reflecting changes
since the last differential abstract block set was created (rather
than forming differential abstract block sets to reflect all
changes since the last full backup). Of course, this method will
work with only one differential abstract block set as well.
At a step 260, the most recent full or differential abstract block
set is selected. Of course, this selection is made from those
logical objects that were recorded before the target restore time
(differential abstract block sets more recent than the target
restore time reflect more recent data than should be restored.) At
a step 261, all of the logical data blocks that are not in the
merged abstract block set are appended to the merged abstract block
set.
Referring to FIG. 25, the first abstract block set selected at step
260 is the differential abstract block set 243. As there are no
blocks in the merged abstract block set yet, the two data blocks of
differential abstract block set 243 are added to the merged
abstract block set 253--corresponding to the first two data blocks
253a-b.
At a step 262, it is determined whether all of the differential and
full abstract block sets have been examined. If not, processing
continues at a step 260.
Returning to the example of FIG. 25, the next abstract block set to
be selected is the full abstract block set 242. At step 261, those
logical data blocks that are already in the merged LBO may be
added. This corresponds to each of the data blocks, other than 242a
and 242b.
At this point, once all of the abstract block sets have been
examined, processing continues at a step 263. At step 263, the
metadata for the merged abstract block set is created. Using the
example of FIG. 25 and 24, the metadata may be of the same
format--the physical address of the logical block elements has not
changed. Accordingly, the metadata is the same. In other
embodiments for formatting metadata, the metadata table may be
updated and correspondence with its format.
The merged abstract block set may be used for copying and restore
in the same manner as an original, level zero abstract block
set.
Primary to Secondary Storage Node Transfers, Example of One
Secondary Storage Node
As described above with respect to FIGS. 11A and 11B, one aspect of
storage systems involves transfer of data from primary storage
elements or nodes to secondary storage elements or nodes.
FIG. 27 illustrates one example of a particularly advantageous
mechanism for transferring data from a primary storage node to a
secondary storage node for storage on tape. This example embodiment
and the components of FIG. 27 are useful both in the context of the
other inventions described above (although not limiting with
respect to those inventions), as well as useful for systems
implemented independent of those inventions.
FIG. 27 includes a primary storage node 270. This may be, for
example, a Symmetrix storage system as described above. In such a
system, a host adapter 270a may be provided for communication with
a host. Disk adapters may provide an interface with the disks. A
remote adapter 270c may handle communications with remote devices,
whether through a SCSI link, an ESCON link, a fiber channel, a
switched network, or some other communication channel. In addition,
a cache 270b may be provided for caching received and transmitted
data.
FIG. 27 also illustrates a secondary storage node 271. In this
embodiment, the secondary storage nodes has a plurality of data
moving elements 271a, 271b, 271e and 271f. In this embodiment, the
data moving elements are arranged in pairs--a front end and back
end pair. For example, data mover 271a may be a front end data
mover--primarily responsible for receiving data from a primary
storage node. The front end data mover 271a may be paired with a
back end data mover 271e. The back end data mover is responsible
for moving data from the secondary storage node to the backup
media.
As shown in FIG. 27, more than one pair of front end and back end
data movers may be provided for parallel transfer of data. In this
example, two pairs are shown--271a-271e, and 271b--271f.
The actual backup media in the example of FIG. 27 is a tape library
272 (other backup media may be used in other embodiments). The tape
library may include a plurality of tape drives 272a-d, each of
which is capable of reading and writing data from a tape (and which
may include an appropriate communications adapter, e.g., a SCSI
adapter). The tape library 272 may also include robotics 271f
capable of selecting tapes from a tape library 272g and inserting
those tapes into the drives 272a-272d. A robotics interface 272c
may control the selection process.
Returning to the secondary storage node 271, the secondary storage
node may include an internal storage device 271c for buffering data
received from the front end data mover (e.g., 271a), before being
written to tape by the back end data mover (e.g., 271e) during a
backup (or, conversely, for buffering data during a restore by
placing the data in the internal memory 271c (by a backbend data
mover 271e) and forwarding the data to a primary storage node (by
front end data mover 271a).
The data movers 271a, 271b, 271e and 271f may be Intel based
personal computers, running software permitting the data movers to
transfer data from the primary storage node to the tape library
unit during backup, and vice versa during a restore.
As described above, the data movers are configured in pairs, e.g.,
front end data mover 271a and back end data mover 271e. Each pair
of data movers may be used to define one or more virtual circuits
or streams.
The front end data mover (e.g., 271a) may be connected to the
primary storage node 270 using any of a variety of connections. For
example, in the example of FIG. 27, two ESCON cables are used to
connect each front end data mover to the ports of a remote adapter
of a single primary storage node (e.g., a Symmetrix storage
device).
In the example of FIG. 27, the back end data movers 271e, 271f are
connected to the tape library unit 272 using SCSI cables. In this
example, each SCSI connection goes to a single read/write drive
272a-272d of the tape library 272. Of course, the SCSI connections
may be daisy chained, permitting more than one drive to be
connected to each back end data mover port. Other connections could
be used, including other links or even a switched network.
The internal storage memory 271c may itself be an iterative cached
disk array, such as a Symmetrix. Thus, a Symmetrix product may be
included as an internal caching memory for movement of data from
the front end to the back end. The internal memory device 271c may
include a service processor, such as a laptop personal computer for
local control of the internal storage device 271c. The internal
storage device may also store the operating system and application
programs running on the data movers 271a, 271b, 271e, 271f and the
control station 271g.
The control station 271g may be an Intel machine, running any of a
number of operating systems, such as SCO UNIX. The control station
271g may also include a keyboard and screen for local operation of
the control station 271g.
The control station 271g controls operation of the data movers
271a, 271b, 271e and 271f. The control station 271g includes
controller software 271b to perform this function. The controller
271b also is used for system configuration and monitoring system
performance. The control station 271g includes a database 271i
(which may, in the alternative, be stored on the internal memory
271c). The database 271i stores information about all pending
backup streams or sessions, the contents of tapes in the tape
library unit and other control information for managing the backup
process and backup media.
The control station 271g may also include an interface 271j for
manipulating and controlling the robotics of 272c, 272f of the tape
library unit 272.
As described above, the primary storage node 270 may be used as the
interface between host connectors (e.g., host computers connected
to host adapter 270a) and secondary storage node, 271. In these
embodiments, and where the storage management application resides
primarily on the host computer, the primary storage node 270 may be
used to pass commands from the host computer to the secondary
storage node 271. Such commands may include instructions directed
to mounting and dismounting tapes, reading and writing tape headers
and trailers and other commands.
The primary storage node 270 may simply pass appropriate commands
to the secondary storage node 271. In the alternative, the primary
storage node 270 may perform some functions based on those
commands, such as format checking.
As described above, the backup restore process can be performed by
establishing a virtual channel between a primary storage node 270
and the tape library 272, through the secondary storage node 271.
As described above, this may involve formulating a connection
through a network between primary storage node 270 and secondary
storage node 271. This may also involve establishing a connection
with a tape drive 272a and applicable tapes 272g.
FIG. 28 illustrates one example of a state diagram for a secondary
storage node, such as node 271, for establishing and maintaining a
virtual channel. At state 280, a backup control stream session (or
virtual channel) is requested by the storage management application
(e.g., on the host computer). Establishment of the virtual channel
may involve selecting an appropriate front end and back end data
mover pair, e.g., front end data mover 271a and back end data mover
271e.
A function to be performed by the storage management application
may require opening a tape. The result would be to place the
secondary storage node 271 into state 281--virtual channel
beginning of tape. This transition would involve mounting the
appropriate tape, using similar techniques to what is known in the
art. At the beginning of tape state 281, tape headers and trailers
may be read or written, as a part of the tape management
process.
When it is time to record information on the tape, the secondary
storage node 271 (or at least the applicable data movers within the
secondary storage node) enter the virtual channel write state 282.
When in this state, the recording part of a backup is performed,
such as writing one or more abstract block sets, or portions of an
abstract block set, to tape.
If the end of a tape is encountered, the applicable data movers in
the secondary storage node 271 enter the virtual channel end of
tape state 284. In this state, the applicable catalog information
may be read and an appropriate tape trailer written. When the end
of the tape is encountered (or end of data), the applicable virtual
channel needs to close that tape, returning the data movers and the
secondary storage node to the initial state when the channel was
formed--state 280.
If an error is encountered, during writing from state 282, the
virtual channel can enter into an error state 283. The tape may be
closed (returning to state 280), an error log created, and a system
operator notified.
As discussed above, the storage management application is
responsible for issuing the appropriate commands to change the
state of the secondary storage node 271. The storage management
application may be resident on the host computer, primary storage
nodes, separate network storage controller or even on the secondary
node 271.
FIG. 29 illustrates a state diagram for the secondary storage node
271 for restoring information from tape. The state diagram begins
at state 291, where a request to open a virtual channel has been
received. The storage management application handles the opening of
tapes, for example by requesting a tape open for the backup channel
stream. This results in entering the virtual channel beginning of
tape state 292. As before, this can include tape header and trailer
reads as well as reading of abstract block set metadata, for
systems using abstract block sets.
The actual reading of data can be controlled using a tape read
command, causing the secondary storage node 271 to enter into the
virtual channel read state 293. At end of tape (or data) or
log-out, the secondary node may return to the virtual channel end
of tape state 292. The tape may then be closed, returning the
secondary storage node 271 to the virtual channel opened state.
If an error is encountered during reading, the node 271 may enter
the error state 294, similar to the error state described above
with reference to FIG. 28. When an error occurs, the tape may be
closed, an error log created, and the system operator notified.
For both backup and restore, the cataloging and identification of
tapes can be handled by the storage management application, as is
done for other mechanisms for formatting data stored on a storage
system. The control station 271g of the secondary storage node 271
assists in identification and mounting and dismounting of the
appropriate tapes, using the control station database 271i.
The backup and restore state diagrams of FIGS. 28 and 29 constitute
example embodiments of placing the system (e.g., the primary
storage node and/or the secondary storage node) in an asynchronous
transfer state. In particular, the nodes of the storage domain
enter a state where data is transferred independent of control from
any host computer or host domain element, even when much of the
storage management application process (and software) is being
performed on the host computer.
Certain embodiments of this facet of the invention allow the
advantage of independent control and transfer of copying, backup
and restore. In certain embodiments of the invention, this can
alleviate the dependence on particular host platforms and conserve
host resources. Certain embodiments of this aspect of the present
invention also allow for increased scalability--allowing addition
of memory, with less dependence on host configuration.
One Embodiment of Data Transfer
FIG. 30 illustrates one embodiment of an architecture for a primary
storage node that facilitates transfer of data to a secondary
storage node or to another primary storage node. This embodiment
(as well as others) may be used to implement one or more of the
above inventions.
FIG. 30 illustrates a primary storage node 300. The primary storage
node 300 includes a remote adapter 301, as generally described
above with reference to FIG. 7. The primary storage 300 also
includes a disk adapter 305, also configured as generally described
above with respect to FIG. 7.
Data is stored among a plurality of disks within the primary
storage node 300, one of which is shown in FIG. 30--disk 306.
The disk 306 may include protection bits, as described above with
reference to FIG. 20. These protection bits may be used to
designate tracks to be copied--and also tracks which should not be
updated before they are copied. The protection bits 307 may be
stored, in one embodiment, on a cylinder header for the disk device
306. The disk device 306 may also include a physical bit mask (not
shown) as generally described above with reference to FIG. 22.
Other mechanisms may be used for marking or recording, which tracks
are protected.
In the embodiment of FIG. 30, the disk adapter 305 receives
instructions from the storage management application as to what
physical backup elements (here, which of the tracks 308a-e) are
part of the backup process. The disk adapter may then write the
protection bits at the time of backup is initiated.
Those physical backup segments (e.g., tracks 308a, 308b and 308e)
that were designated as part of a backup process may then be copied
to a side file 303 in a cache 302 of the primary storage node 300.
Thus, the side file 303 may receive the designated tracks 308a,
308b and 308e for copying to another storage node. The side file,
therefore, may contain copies 303a-c of these tracks.
In addition, the disk adapter 305 may post, to a request queue, a
request that the physical backup segments that have been copied to
the side file 303 be transferred to another node. Thus, requests
304a-c may be posted in the request queue 304, corresponding to
those physical backup segments in the side file 303.
The remote adapter 301 may pickup requests from the queue and
transfer copies of the applicable track to the receiving storage
node, e.g., a secondary storage node.
The applicable storage backup segments held in the side file 303
may be part of more than one copy of backup process being
performed. For example, more than one abstract block set may be in
the process of being backed up over more than one virtual channel
connected to the remote adapter 301. In this case, the applicable
metadata for the abstract block set can be used to identify a
specific abstract block set and virtual channel for the copying or
backup process.
In an alternative embodiment, the receiving storage node may
classify physical backup segments based on the abstract block set
to which they belong. For example, the front end data movers
described above could receive physical backup segments
corresponding to tracks, including a physical address for the
track. The front end data move may be aware of the metadata for the
abstract block set, which was formulated by the storage management
application (which identified all of the physical locations for the
applicable logical object being backed up). This would permit the
front end data mover to classify the physical backup segment based
on its physical address.
Of course, a variety of alternative structures and methods could be
employed for transfer through a side file. As just one example, the
physical backup segments could be sorted into separate side files
for each abstract block set (or other structure) being copied or
backed up. In addition, side files may be used to accumulate
segments of data for transfer. For example, a side file could be
created that includes at least ten megabits of data before transfer
through the remote adapter 301 to a secondary, or other, storage
node.
FIG. 31 illustrates one embodiment of a method for using the
structure shown in FIG. 30. At a step 310, the protection bits (307
of FIG. 30) are marked for physical backup segments being copied.
As described above, this may include marking the bits for more than
one logical backup object.
In addition, metadata for the applicable logical object may be
transferred to the receiving storage node, e.g., the secondary
storage node. Thus, if the metadata is of the form shown at 133 of
FIG. 13, this metadata may be specified and advance the backup
process. This metadata may (or may not) be reformulated during
backup for incorporation into the logical backup object, such as
reformulation into the form shown at 134 of FIG. 13. In any event,
this metadata may be used by the disk adapter 305, remote adapter
301 and/or the receiving storage node to accumulate and organize
the applicable physical segments associated with the logical object
being copied or backed up.
At a step 311, the protected segments are transferred to a side
file in a cache. As this is done, requests for the transfer of the
physical backup segments are logged into a request queue. As
described above, this may be performed by a disk adapter of the
primary storage node. At this point in time, the disk adapter 305
may also reset the applicable protection bit of the protection bits
307 of the disk device 306, allowing future updates of the
data.
The segments in the side file can then be transferred to another
storage node by the remote adapter 301, such as transfer to a
secondary storage node. This may be done be reading requests for
transfer from the requests queue 304.
After the transfer (e.g., after the transfer has been acknowledged)
the applicable entries for the segment in the request queue in the
side file may be removed. Of course, this can simply be done by
allocating the storage as unused.
FIG. 32 illustrates one example of data flow in a backup process
through a secondary storage node 320. In this embodiment, the data
is initially received by front end processor 322. The front end
processor may be as generally described above with reference to
FIG. 27.
The front end processor 322 stores the received physical backup
segments in internal memory 323 in files associated with the
applicable entity being backed up. For example, if an abstract
block set LBO #1 is being backed up, the physical segments are
stored in a file 324 associated with that abstract block set. Where
more than one abstract block set is being transmitted at the same
time over a virtual channel, the front end processor may sort the
applicable physical data segments into the appropriate file, e.g.,
files 324 and 325.
When a file reaches a certain threshold size, for example 10
megabits, the front end processor 322 may notify the back end
processor 326 that a segment of the abstract block set is ready for
copying to tape. The back end data mover 326 may then copy that
portion of the abstract block set from the internal memory 323 to
the tape library unit 321.
In the event that the internal memory 322 is an iterative cached
disk array, such as a Symmetrix, the physical back up segments may
be copied from the applicable file 324, 325 by the back end data
mover 326 in last-in-first-out order. This may be done to increase
the chance that the data is copied from a cache within the internal
memory 323, rather than from disk within the internal memory
323.
As described above, more than one abstract block set may be backed
up at one point in time over a virtual channel. In addition, the
segments of an abstract block set may be written in fixed sized
pieces. For example, if an accumulation file 324, 325 accumulates
physical back up segments until a threshold size (for example 10
meg) is reached, the abstract block set may be stored in
interleaved segments of a tape. The controller and control station
(271h and 271g) can maintain a database for this information. In
addition, the applicable information can be written to appropriate
headers and trailers on the tape.
FIG. 33 illustrates one example of a tape containing backup
information written by a device according to one embodiment of the
present invention.
In FIG. 33 the tape has a beginning portion 330 and an ending
portion 332. The beginning portion 330 includes the usual tape
header 330a, and perhaps a specific tape header for the secondary
storage node 330b. After the tape headers 330a, 330b, the tape
includes interleaved segments of abstract block sets (including
metadata) 338, separated with file marks. For example, the
interleaved segments may include a record 331 that includes a
series of copies of physical backup segments 331b. A segment header
331a and segment trailer 331c may identify and separate this
portion of the abstract block set from other portions of the
tape.
Interleaved with the portion of the abstract block set that
includes data blocks 331 may be other abstract block set physical
backup segment records for this and other abstract block sets. In
addition, a record 333 may be written that includes abstract block
set metadata. This metadata 333a may be of any of the forms
described above, or other formats. As a part of the segment header
information 331a and 333a, the applicable abstract block set can be
identified, for example, using an abstract block set identifier
uniquely assigned to each abstract block set. This permits
identification and coordination of the records interleaved on the
applicable tape 330, 332.
At the end of the tape 332, a tape directory 334 may be written.
Similarly, server tape trailer information 335 may be written. At
the end of the tape, a tape catalog 336 and a secondary storage
device tape trailer marking the end of the tape 337 may be
written.
Using a database of tapes, the applicable information may be
retrieved from a backup tape. Because abstract block sets may
include data blocks written in any order, a restore process can
efficiently retrieve and write the portions of an abstract block
set being restored, in any order. This permits the storage
management application to identify each of the tapes that include
portions of an abstract block set and to mount (and read all of the
applicable portions of) those tapes only once. Of course the first
tape to be mounted may be the tape that includes the metadata
records for the abstract block set being restored. For this reason,
it may also be preferable to record the metadata at one end of all
of the segments of an abstract block set written on the tape
holding the metadata--making the reading of metadata at the
beginning process simpler. This permits formation of the
appropriate mapping table, described above, for the restoration
process to proceed independent of the order in which data blocks
are retrieved.
For the reasons described above, the reading and restoring of data
blocks within an abstract block set can be done in any order. As a
result, where tapes are used and as a component of the secondary
storage element, the tapes can be mounted and dismounted in any
order for both storing and retrieving data. As a result, where more
than one tape drive is present in the secondary storage element, it
is shown in the embodiments described above, data blocks can be
written during backup and read during restore and parallel using
multiple drives.
Referring to FIG. 14, parallel writing of data may be performed as
followed. In this example, the updating of metadata (step 147) may
be performed entirely in advance. In this example, the metadata may
be the physical addresses of the data being read in a primary
storage element. Accordingly, all of the metadata can be determined
in advance of the actual backup process. Since this is the case,
the steps 144, 146 and 147 may be performed in parallel. That is,
after the physical backup segments have all been identified and the
metadata determined (e.g., at step 142), all of the data blocks may
be read in parallel and written to multiple tapes in parallel. In
one embodiment, the last tape (which may be randomly selected) can
store the metadata at the end of all of the data blocks that are
part of the abstract block set on that tape.
One example of a parallel restore operation may be described with
reference to FIG. 15. As described above, at steps 150-153, the
mapping for the restore of the logical object is determined. Where
this restore is coming from a tape, the metadata for the abstract
block set can be retrieved in advance. As described above, after
this has been done, the abstract block sets can be restored in any
order. Accordingly, the abstract block sets may also be retrieved
in parallel using multiple tape drives for a restore. In this case,
the steps 154-158 may be performed in parallel using multiple tapes
(or other media) for retrieving data blocks of the abstract block
set being restored.
In embodiments employing virtual channels, a separate virtual
channel may be established for each of the parallel paths for
transfer o f data. For example, a separate virtual channel may be
established for each tape drive. In another embodiment, a single
virtual channel may be established, but permitting multiple tape
drives to channel data into that virtual channel. This may be
particularly advantageous where the speed of reading data from the
tape drive is slower than the ability to transfer data from the
secondary storage node to a primary storage node. Allowing parallel
reading of tape drives permits the speed of the restore to approach
the ability of the connections to transfer data and the primary
storage element to receive that data.
While many of the above embodiments have been described with
respect to backup and restore operations between a primary storage
element and a secondary storage element, many aspects of the
invention have much broader application. As just one example, an
abstract block set can be used for any transfer of data. As another
example, the application of a secondary storage node can be greater
than simply backup and restore operations. Such storage nodes may
also be used for hierarchical storage management applications,
operation of virtual disks, and other applications.
The various methods above may be implemented as software on a
floppy disk, compact disk, or other storage device, for use in
programming or controling a computer. The computer may be a general
purpose computer such as a work station, main frame or personal
computer, that performs the steps of the disclosed processes or
implements equivalents to the disclosed block diagrams. The
software may be included on a diskette as a complete system or as
enhancements to an exisitng system, permitting the system to
perform the methods described herein.
Having thus described at least illustrative embodiments of the
invention, various modifications and improvements will readily
occur to those skilled in the art and are intended to be within the
scope of the invention. Accordingly, the foregoing description is
by way of example only and is not intended as limiting. The
invention is limited only as defined in the following claims and
the equivalents thereto.
* * * * *